advances in electromagnetic fields in living systems

ADVANCES IN

ELECTROMAGNETIC FIELDS

IN LIVING SYSTEMS

Volume 3

ADVANCES IN ELECTROMAGNETIC FIELDS

IN LIVING SYSTEMS

Volume 3

Edited by

James C. Lin University of Illinois at Chicago

Chicago. Illinois

Springer Science+Business Media, LLC

The Library of Congress cataloged the fiest volume of this title as follows:

Advances in electromagnetic fields in Iiving systems I edited by James C. Lin. p. cm.

Includes bibliographicaI references and index. 1. Electromagnetic fields-PhysiologicaI effect. 2. Electromagnetic fields-Health aspects.

3. Electromagnetic fields-Therapeutic use. 1. Lin, James C. QP82.2.E43A29 1994 591.19'17-dc20 94-24263

ISBN 978-1-4613-6886-1 ISBN 978-1-4615-4203-2 (eBook) DOI 10.1007/978-1-4615-4203-2

©2000 Springer Science+Business Media New York OriginaIly published by Kluwer Academic / Plenum Publishees in 2000 Softcover reprint of the hardcover 1 st edition 2000

http://www.wkap.nl!

10 9 8 7 6 5 4 3 2 1

A c.I.P. record for this book is available from the Library of Congress

Ali rights reserved

CIP

No part of this book may be reproduced, stored in a retrievaI system, or transmitted in any form or by any means, electronic, mechanicaI, photocopying, microfilming, recording, or otherwise, without written

permission from the Publisher

PREFACE TO VOLUME 1

Hardly any phenomenon in the modern environment is as ubiquitous as electromagnetic fields and waves. We have learned to understand the physical characteristics of these energy forms, and we have applied them in abundant ways to embellish our ways of life and our standards of living. Furthermore, we have come to depend on them for health, safety, information, comfort, and conveyance.

Apart from their intended roles, these electromagnetic fields and waves produce other effects that may influence the activities of living organisms. The effects produced depend on many physical, chemieal, and biological factors. They may be grossly apparent and visible soon after exposure of the living organism or they may not appear to have influenced the organism at all upon casual examination. Even then, there may be subtle changes that are only detectable upon careful chemical or microscopic study, or which are apparent only after a considerable time delay.

Nevertheless, our understanding of the interaction of electromagnetic fields with living systems is advancing in a wide range of topical areas. This bi-annual series with invited reviews by recognized leaders in their respective specialties, will present progress to date in key areas of research and scholarship. The guiding philosophy of this undertaking is the presentation of integrated, known, and confirmed phenomenological observations, basic mechanisms of interactions, and applications in biology ilnd medicine, as well as perspectives on current topics of interest.

A further intent of this series is to promote the interchange of ideas between biomedical, engineering, and physical science specialties, and thus the series is aimed at both practitioners and researchers. As the numbers of publications multiply, it becomes a challenge to locate and especially digest such a collection of papers, contradictions, and interpretations. It is hoped that volumes in this series will provide the catalyst in efforts to perfect our knowledge for health protection, and to develop new and better diagnostic, as well as therapeutic, procedures.

In the last two decades, research on the biological effects and health implications of electromagnetic fields not only has expanded, but also has become a subject of a public concern and private debate, worldwide. This series is aimed at bringing together contemporary advances in key areas of research and scholarship. Very seldom can advances be totally divorced from past accomplishments. Accordingly, this premier volume begins with a chapter that discusses, briefly, contributions made by some of the early investigators on the interaction of electromagnetic fields with living systems. The interaction of radiofrequency (RF) electromagnetic fields with the central nervous system has been a subject of considerable contemporary interest, since the nervous system integrates and regulates an organism's response to its environment. The chapter that follows summarizes the known effects of RF radiation on the

v

central nervous system and includes a review of interaction of RF exposure with psychoactive drugs on animals. The latter has become one of the most intriguing research subjects with profound implications for health effects and safety protection.

This volume presents basic, applied, and clinical information ranging from extremely low to super high frequencies. In general, the interaction of electromagnetic fields and waves with biological systems is frequency-dependent. Moreover. the mechanisms of interaction for fields at low frequencies are very different from those at high frequencies. While significant advances are being made on many fronts. a particular emphasis of this volume has been placed on the recent developments of our understanding in the very low and extremely low frequencies. They have led to some exciting new applications in the clinical management of tissue injury from exposure to extremely low frequency electric fields, and in the therapeutic treatment of.the musculoskeletal systems such as the repair of soft tissue wounds and bone fractures using low frequency fields (see Chapters 3 and 4). These low frequency fields also are emitted by circuitry within video display terminals (VDTs) and video display units (VDUs). Recently, considerable concern has been expressed by the public and some professionals about the possible health effects resulting from extended use of these devices. Chapter 5 examines, in detail, the possible effects of electric and magnetic fields from VDTs and VDUs on human health.

It is my pleasure to acknowledge the National Science Foundation, Office of Naval Research, and the National Institutes of Health, for their support of my research throughout the years. Some of the results are found in various parts of this volume. A portion of the writing and editorial activities was completed while on sabbatical leave of absence from the University of Illinois. I am grateful to the National Science Council of ROC on Taiwan, for the opportunity to serve as an NSC Chair Professor during the 1993-94 academic year. Many individuals have contributed to the realization of this volume. I want to thank especially, Mr. L. S. Marchand, the senior editor at Plenum, for his encouragement and patience through all phases of this work. I wish to express my appreciations to the authors for their friendship and willingness to share their intellectual accomplishments. And lastly, but importantly, to my family: my wife, Mei, and children, Janet, Theodore and Erik. I am deeply indebted for their faith. love, patience, and ungrudging support.

vi

James C. Lin Chicago and Taiwan

PREFACE TO VOLUME 2

This is the second volume in the series on Advances in Electromagnetic Fields in Living Systems. The objective ofthis volume is to add to thescientific and professional literature a number of significant pieces of research larger in scope than journal articles. We hope that this form of publication will make the information readily available to research organizations, libraries, government agencies, independent investigators, and interested persons.

The chapters in this volume are organized into two consecutive sets using two specific regions of the electromagnetic spectrum: extremely low frequency fields and radiofrequency radiation. While significant advances are being made on both fronts, greater emphasis of this volume is placed on recent developments at radio frequencies. Each chapter consists of a comprehensive review of a topic of current interest and growing importance. Much of the information is based on authors' own research and that of the contributions from investigators in the relevant scientific disciplines.

The first two chapters of the book review two of the most significant topics that have played pivotal roles in raising and addressing the question of whether extremely low frequency (ELF) electric and magnetic fields can affect the development of cancer. Chapter I scrutinizes the connection between exposure to ELF electric and magnetic fields and melatonin synthesis or utilization. It examines data that have been reported to indicate that exposure of animals to ELF fields reduces the ability of these animals to produce this hormone. And it discusses the significance of the findings relative to the incidence of cancer in humans exposed to ELF fields. The large number of epidemiological reports that focus on cancer and its potential association with ELF exposure are evaluated in Chapter 2. It provides a strength evaluation for the available evidence at this time and a discussion on the unique challenges that face epidemiological studies of ELF exposure.

An important task in assessing health risk from exposure to ELF and radiofrequency (RF) electromagnetic fields is the quantitative determination of ELF and RF fields within and without biological bodies. The emphasis of Chapter 3 is on computational methods for dosimetry and exposure assessment and their application in bioelectromagnetic investigations. It provides a general knowledge base for ,computational bioelectromagnetics. It also gives specific guides to computing ELF and RF coupling and field distributions inside homogeneous and nonhomogeneous phantom and animal bodies.

The biological effects of RF and microwave radiation have become a focal point of attention because of the accelerated use of RF radiation for wireless communication over the past few years. Wireless communication systems use low power modulated forms of RF and microwave radiation that were not investigated extensively in the past. Research addressing issues pertaining to the wireless communication spectra has begun only recently. Chapter 4

vii

summarizes results from published studies using frequencies in the same spectral band and provides information on current research activity. It includes carcinogenesis and cancer promotion by RF and microwave exposure, and other in vitro and in vivo experimental studies that involve primarily the central nervous system and other tissues in the head. A brief description of epidemiological studies on RF and microwave exposure is also inlcuded. The material should be of use for preliminary risk assessment.

Chapter 5 examines, in detail, the reported experimental evidence for possible effects of RF fields on cancer initiation, promotion, and progression. It provides a necessary background for the direction of future laboratory research to help clarify whether RF and microwave radiation influences cancer initiation and development. It examines the critical parameters of the exposure that may account for any influence. An exciting new medical application, the clinical management of cardiac arrhythmia using catheter-delivered RF and microwave energy, is summarized in Chapter 6. RF cardiac ablation has become the most commonly used minimally invasive procedure for treatment of irregular heart rhythm. Microwave energy is a viable alternative energy source for percutaneous catheter ablation (additional references to this energy source are give at the end of Chapter 4).

While the health effects ofRF and microwave radiation remain a concern to the general public and many professionals, the new ANSIIIEEE exposure standard represents a wealth of scientific understanding and significant improvement over its predecessor. However, its complexity has caused difficulties in the implementation of the standard in the real world exposure situation. Chapter 7 provides guidance on what is involved in assessing exposure and offers insights to applying the standard from a practical perspective.

Lastly, I wish to thank the authors for their important contributions. I also want to pay a special tribute to the investigators in this field. whose published works and personal communications greatly helped us in writing the chapters. As always. lowe a huge dose of gratitude to my family for their faith and support.

viii

James C. Lin Chicago

PREFACE TO VOLUME 3

The past few years have been exceptionally active periods for research on the interaction of electromagnetic fields with living systems. The subject has become a focus of attention because of the expansion of electric power use and distribution at 50 and 60 Hz in the extremely low frequency (ELF) spectrum between 3 Hz and 3kHz and because of the accelerated use of radio frequency (RF) radiation (300 MHz to 6 GHz) for wireless devices over the past decade. In addition to the primary intended roles, electromagnetic energy may produce effects that could influence the vital activities of living organisms. A major research effort related to possible adverse health effects of environmental electric and magnetic fields (EMF) was completed in the United States. A report, summarizing the research, was issued by the administering agency, the National Institute of Environmental Health Sciences [NIEHS, 1999]. Another report, appraising the research, was issued by the National Academy of Sciences [NASfNRC, 1999].

In. the summary report, NIEHS concluded that "the scientific evidence suggesting that ELF-EMF exposures pose any health risk is weak. The strongest evidence for health effects comes from associations obse~ved in human populations with two forms of cancer: childhood leukemia and chronic lymphocytic leukemia in occupationally exposed adults. While the support from individual studies is weak, the epidemiological studies demonstrate, for some methods of measuring exposure, a fairly consistent pattern of a small, increased risk with increasing exposure that is somewhat weaker for chronic lymphocytic leukemia t.han for childhood leukemia. In contrast, the mechanistic studies and the animal toxicology literature fail to demonstrate any consistent pattern across studies although sporadic findings of biological effects (including increased cancers in animals) have been reported. No indication of increased leukemia in experimental animals has been observed."

The National Academy of Sciences' review and evaluation concluded that results, from the research that was authorized by Congress, "do not support the contention that the use of electricity poses a major unrecognized public health danger." Moreover, the report stated that, "The biologic research contributed little evidence to support the hypothesis that a link exists between MF (magnetic field) and cancer." However, the report went on to state that, "In contrast with laboratory research, some epidemiological studies have reported difference in incidences of cancer associated with MF exposure that differs by as little as 0.2 to 0.4 J.L T." It is interesting to note that these two reports reached a similar conclusion-there is a need for further research, especially at the basic science level.

This volume begins with a chapter describing the difference in coupling of ELF fields and RF radiation into biological systems and mechanisms of interaction disclosed through the Maxwellian formulation. While ELF fields and RF radiation occupy the same known electro-

ix

magnetic spectrum, their mode of coupling into biological tissues and their mechanism of interaction can be very different. It is shown that electromagnetic energy at ELF and RF frequencies can be used to probe the body in different ways. These observations suggest that it is possible to deduce explanations for a broad range of biological interactions and reactions through an examination of the coupling mechanisms for ELF fields and RF radiation. The next four chapters present current and future biomedical applications of electromagnetic fields and RF radiation in diagnostic imaging and therapeutic treatment- The last two chapters are devoted to the biological effects of pulse RF radiation.

Clinical evaluation of magnetic resonance imaging (MRI), which relies on constant and RF magnetic fields for its operation, had begun in the 1980s. Continued developments in instrumentation and applications since then have allowed MR! to become the diagnostic imaging modality of choice in many medical practices. Moreover, the dramatically reduced scanning time has accelerated and diversified medical applications of MRI in recent years. Applications that are curr~ntly under investigation include magnetic resonance angiography, perfusion and diffusion imaging, temperature mapping, and functional imaging. Chapter 2 reviews some of the techniques and applications of MR!, and introduces a novel tissue impedance MR! technique for the brain. Imaging electrical activity of the brain is a promising area of noninvasive biomedical application of low frequency electric field. Chapter 3 provides a state-of-the-art review of imaging brain electrical activity using scalp electroencephalograms (EEG). The chapter discusses some of the inverse imaging algorithms used for EEG spatial deconvolution imaging, namely cortical-potential imaging in a realistically shaped inhomogeneous head model.

The use of low frequency electric fields to improve and help maintain functions of the living system is a dazzling topic. External electric fields applied to biological materials interact directly on free electric charges and dipoles. They may cause intermolecular transitions and intermolecular processes than can lead to structural reorganization of the cell membrane through the technique of electroporation and electropermeabiIization. The induced transient changes in membrane states have led to electric field pulse techniques to gain increasing importance in cellular and molecular biology. in gene technology. and in therapeutic medicine. Chapter 4 gives a detailed account of the electric field pulse techniques and in vivo applications of electropermeabilization in combination with chemotherapy or as a tool for gene transfection. Likewise. a better understanding of the fundamental mechanisms of electric field induced tissue modification and injury could help to develop efficient therapeutic procedures and effective management tactics. Chapter 5 discusses the theory and experimental evidence concerning the effect of intense electric field on cell membrane proteins and membrane active transporters.

The fact that pulsed RF radiation was used almost exclusively for radars meant that few people in the general population would have encountered exposure to pulsed RF radiation as a by-product of technological aids in their daily activity. The science and engineering of RF technology have expanded to the point that many types of devices have been developed using pulse RF energy to embellish our lives. Indeed. we have come to depend on the pulse RF energy for health, security, information, and entertainment. Moreover, recent developments in high peak power pulsed microwave systems have renewed interest in exploring potential biological effects of high peak power RF pulses. Currently, there are systems capable of delivering 100 gigawatt (GW) pulses to a transmitting antenna and establishing hundreds of kV/m peak E-field intensity in the beam path. However, we know very little about the safety aspects of high peak power, ultra-short RF signals. Chapter 6 preserUs an up-to-date review of the biological data available in the English language for pulsed RF fields.

x

It has been recognized for years that significant contributions to the field of bioelectromagnetics have been made by research performed in the former Soviet Union (FSU). Unfortunately, most of this research was published in the Russian language and was not readily available to scientists in the West. As a result, these publications have not been reviewed comprehensively in English until now. Chapter 7 gives a comprehensive review of the FSU research on biological effects and potential health hazards of pulse RF radiation. Most of the reported effects of low-intensity pulsed microwaves were subtle functional changes, which did not exceed the limits of normal physiological variation and could only be detected by sensitive physiological tests. However, some studies did report clearly pathogenic effects. Interestingly, such issues as RF-induced carcinogenesis apparently had not been a concern and were not studied at all. Nevertheless, some of the findings may have implications on the conceptual understanding of interaction mechanims and on approaches to RF safety protection. Their replication by researchers in the West deserves consideration.

And, finally, to the authors who have generously contributed their expertise, kindly developed their chapters, and patiently endured several stages of change, I am sincerely grateful. It is my earnest hope that the information presented in this volume will inspire scientific investigators to continue their efforts to expand our knowledge of the interaction between electromagnetic fields and living systems.

REFERENCES

James C. Lin Chicago

NAS/NRC, 1999. Research on Power Frequency Fields Completed under the Energy Policy Act of 1992, National Academy Press. Washington, DC.

NIEHS. 1999. Health Effects from Exposure /0 Power-Line Frequency Electric and Magnetic Fields, Prepared in Response to the 1992 Energy Policy Act (PL \02-486, Section 21\8) (NIH Publication 99-4493), Washington, DC.

xi

CONTENTS

Mechanisms of Electromagnetic Field Coupling into Biological Systems at ELF and RF Frequencies .......................................................................... 1 JamesC. Lin

Principles and Horizons for Magnetic Resonance Imaging .......................................................... 39 Shoogo Ueno and Norio Iriguchi

Imaging Brain Electrical Activity ................................................................................................. 73 Bin He, Dezhong Yao, and Dongsheng Wu

Applications and Control of High Voltage Pulse Delivery for Tumor Therapy and Gene Therapy in vivo ' ........................................................................................... 121 BertH R. R. Persson

The Electric Field-induced Electroconformational Coupling of Cell Membrane Proteins ......... 147 Wei Chen

Biological Effects of High Peak Power Radiofrequency Pulses ................................................. 207 Shin-Tsu Lu and John O. de Lorge

A Comprehensive Review of the Research on Biological Effects of Pulsed Radiofrequency Radiation in Russia and the Former Soviet Union ......................................... .265

Andrei G. Pakhomov and Michael R. Murphy

Contributors ............................................................................................................................ 291

Index ........................................................................................................................................ 293

xiii

MECHANISMS OF ELECTROMAGNETIC FIELD COUPLING INTO BIOLOGICAL SYSTEMS AT ELF AND RF FREQUENCIES

JamesC. Lin

Electrical Engineering and Computer Science and Bioengineering Departments

University oflllinois at Chicago Chicago, IL 60607-7053

INTRODUCTION

Although living organisms have thrived in a natural electromagnetic environment, increasingly, they have been subjected to a myriad of man-made electromagnetic fields and radiations. Besides the primary intended roles, these fields and waves produce other effects that may influence the vital activities of these organisms. The changes produced depend on many physical and biological factors. They mayor may not be grossly apparent and be visible soon after exposure of the living organism.

Electromagnetic energy in the frequency region below 300 GHz has wavelengths in air longer than 1 mm. Furthermore, at wavelengths closer to the millimeter limit of the spectrum, electromagnetic energy behaves as infrared radiation. They produce photons of low energy, therefore, under ordinary circumstances, they are too low to produce ionization or excitation. Consequently, they are often referred to as low energy or nonionizing radiation. Electromagnetic energies propagate or progress through a material medium at a constant speed in that medium. In particular, it propagates through air or vacuum at the speed oflight, 2.998 x 108 mls. Energies with wavelengths longer than 10 m (lower than 30 MHz) have interaction properties that differ greatly from those of wavelengths that approximate the human body's physical dimensions.

It is customary for telecommunication use to divide the spectrum into bands with specific designations. Since its interaction with biological media differs according to the specific spectral band, these properties can give rise to different effecl<; and applications. For example, the radiofrequency (RF) band of300 kHz to 30,000 kHz (or 0.3 to 30 MHz with wavelengths from 1000 m to 10m) is used in medicine for ablating, coagulating and cauterizing tissue. Frequencies lower than 10kHz is avoided to prevent stimulation of excitable muscular and cardiac tissues [Lin, 1999].

Advances in Electromagnetic Fields in Uving Systems, Vol. 3 Edited by J.C. Lin, Kluwer AcademicIPlenum Publishers, 2000

The biological effects of electromagnetic fields and waves have been a subject of scientific research for more than 100 years. And, our knowledge regarding the health effects has increased accordingly. Nevertheless, they have become a focus of attention because of the expansion of electric power use and distribution at 50 and 60 Hz in the extremely low frequency spectrum (ELF - between 3 Hz and 3kHz) and because of the accelerated use ofRF radiation (300 MHz to 6 GHz) for wireless communication over the past decade. A notable cause for the attention is the uncertainty and lack of understanding of the mechanism of interaction of electromagnetic fields and waves with biological systems. While ELF fields and RF radiation all are parts of the same known electromagnetic spectrum, their mode of coupling into biological tissues and their mechanism of interaction can be quite different. This chapter discusses the coupling of ELF field and RF radiation and mechanisms of interaction availed through the Maxwellian formulation. For descriptions of the biological effects and medical applications of ELF field and RF radiation. the interested reader is referred to other chapters in the present volume and volumes one and two of this book series.

MAXWELLIAN FORMULATION

To appreciate the interaction of electromagnetic fields and radiations with biological systems, one begins with the physical laws describing their characteristics and behavior in biological media. The mathematical expressions defining this process are known as Maxwell equations. These equations capture our understanding of basic electromagnetic process and theorize mechanisms of interaction through experimental observations. They provide a set of principles for working with abstract ideas and for the logical deduction of all electromagnetic phenomena.

Specifically, electromagnetic radiation consists of electric and magnetic fields that change with space and time. Their spatial variation is dictated by the electromagnetic c properties of tissue media, i.e., electrical permittivity and magnetic penneability. Maxwell equations describe the behavior of electric and magnetic fields in material media distinguished by permittivity and permeability. These equations are valid for linear o' nonlinear, isotropic or anisotropic, homogeneous or inhomogeneous medium in the frequency range from zero to the highest microwave frequencies, including many phenomena at optical wavelengths. Maxwell equations are laws that define the relationship between spatially and temPQrally averaged electric and magnetic fields. They are relevant in regions or volumes whose dimensions are larger than atomic dimensions. The time intervals of observation are typically long enough to allow for an averaging of atomic fluctuations.

Maxwell Equations

The four Maxwell's equations may be specified in either integral or differential form. They will be given in the integral form since they lend to simple physical interpretations. The differential equations may be derived by using the divergence and Stokes theorems from vector analysis.

2

where

E = electric field strength (volt/meter) H = magnetic field strength (ampere/meter) D = electric flux density (coulomb/square meter) B = magnetic flux density (weber/square meter) J = conduction current density (ampere/square meter) p = electric flux density (coulomb/cubic meter).

(1)

(2)

(3)

(4)

It can be seen from the right-hand side of equations (2) and (3) that the resources of electromagnetic fields and waves are charges and current. The auxiliary equations that bridge the fields and flux densities produced by a given current or charge distribution are:

D=eE (5)

(6)

J=oE (7)

Free space or vacuum is a medium in which the permittivity

e = eo = 8.854 x 10-12 = 1I(361t) X 10-9 (F/m), (8)

permeability

~ = ~o = 41t X 10-7 (HIm), (9)

and the conductivity 0 = O. For all other linear, isotropic, and homogeneous media, it is conventional to introduce the dimensionless ratios,

3

(10)

~r= W~O (11)

which are respectively referred to as the relative dielectric constant and relative permeability. Living matters generally have relative permeability close to that of free space and relative dielectric constants that show characteristic dependence on frequency.

According to Faraday's law (1) the total voltage induced in an arbitrary-closed path is equal to the time rate of change of magnetic flux through the area bounded by the closed path. Therefore, a time-varying magnetic field generates an electric field. There is no restriction on the nature of the medium.

Amperes' law (2) states that the line integral of the magnetic field strength around a closed path is equal to the total current enclosed by the path. The total current may consist of two

parts: a conduction current with density J and a displacerr,ent current with density aD/at. Thus,

Ampere's law implies that a magnetic field can only he produced by the flow of current or movement of charges. The displacement current was introduced by James Clerk Maxwell to join the separate laws that govern electricity and magnetism into a unified electromagnetic theory. It also led to the postulate of electromagnetic waves that can transport energy and the knowledge that light is an electromagnetic wave.

Equation (3) is Gauss electric law, which says that the net outward flow of electric flux through a closed surface is equal to the charge contained in the volume enclosed by the surface. Likewise, Gauss law for the magnetic field (Equation 4) states that the net outward flow of magnetic flux through a closed surface is zero. Therefore, magnetic flux lines are always continuous and form closed loops.

Also, the electromagnetic theory requires that at a boundary surface where the tissue permittivity changes abruptly certain boundary conditions must be satisfied. These conditions may be derived by applying Maxwell's equations (1) to (4) to infinitesimal regions containing these interfaces. These boundary conditions maybe swnmarized as follows:

I. The tangential components of the electric field strength are continuous across the boundary.

2. The normal components of the electric flux densities may differ by an amount equal to the surface charge density. The surface charge on the boundary is zero for dielectric materials.

3. The tangential components of the magnetic field strength may differ by an amount equal to the surface current density. In all cases except for perfect conductors, the surface current density is vanishingly small.

4. The normal components of the magnetic flux density are continuous across the boundary.

TISSUE ELECTROMAGNETIC PROPERTY

The interaction of electromagnetic field and radiation with biological systems is characterized by the electromagnetic properties of tissue media, specifically, permittivity and

4

penneability according to the electromagnetic theory outlined by the four Maxwell's equations.

Biological materials have penneability values close to that offree space and are independent of

frequency. It is noted however, that biological tissues are not free of ferromagnetic materials.

The pennittivity of biological materials displays unique dependence on frequency (Fig. 1).

Typically, dielectric constants decrease while conductivities increase with increasing frequency.

Biological materials exhibit very high dielectric constants, especially at low frequencies,

I.:ompared to other homogeneous solids and liquids. This occurs because biological tissues are

composed of macromolecules, cells, and other membrane-bound substances. Mobile counter

ions associated with fixed charges on cell membranes and membrane capacitance dominates the

behavior of dielectric constant at low frequencies. This frequency dependence is the result of the

dramatic change membrane capacitance undergoes as the frequency increases at extremely low

frequencies « 3 kHz). An applied electric field causes charges to accumulate at boundaries separating tissue regions of different dielectric properties, say intra-and extra-cellular spaces.

As the frequency increases, insufficient time is allowed during each cycle to pennit

complete charging of the cell membranes. The total charges per cycle must decrease, along with

the membrane capacitance, as the frequency increases. This behavior gives rise to a decrease in

the dielectric constant between 10kHz and 30 MHz. As the frequency increases still further, the

membrane capacitance change stabilizes until the rotational and vibrational properties of polar

molecules in water becomes significant.

The conductivity of biological tissues behaves in a similar manner. Cell membranes have

a relatively high capacitance at low frequencies. They become progressively short-circuited for

frequencies above 1 kHz, facilitating the intracellular fluid to participate in current conduction.

This causes the conductivity to increase as the frequency increases. At higher frequencies, the

rotation of water molecules is accompanied by viscous loss that accounts for-the principal

mechanism of increased conductivity.

ELECTROMAGNETIC SOURCE FIELDS

Electromagnetic energy at both ELF and RF frequencies can be radiated into space or a

material medium through the use of antennas or radiators that primarily serves as a transition

between the generating source of electromagnetic energy and space or material medium. The spatial distribution of electromagnetic energy from an antenna is directional and varies with

distance from the antenna. At distances sufficiently far from an antenna so that the distribution

changes ()nly with distance, not angle or orientation, the region is called a far field or radiation

zone. At lesser distances, the energy distribution in the near field or zone is a function of both

angle and distance. Moreover, the behavior of electromagnetic fields and their coupling and

interaction mechanisms with biological systems in the near or far zone are very different. In fact,

these differences constitute the major variances between ELF and RF energy deposition into

biological systems. As shown in subsequent sections, the deposition of electromagnetic power,

induction of electric and magnetic fields, absorption energy, and their penetration into tissue are

all functions of the source and its frequency or wavelength. In general, when considering the interaction of electromagnetic fields with biological systems, it is necessary to account for the

frequency or wavelength and its relationship to the physical dimensions of the body. The elementary dipole antenna is important for both practical and theoretical reasons. For

5

1

, , , ~,,,

,.-

1

,

, , , , , ,---~

, , , , #----

",---' ,-_ ... ,

,

Frequency (Hz)

, , , , , I , ,

I I ,

(J

, .. ... -

10-2

Figure 1. Permittivity (dielectric constant and conductivity) of biological tissues (musclt as a function of frequency_

6

instance, any linear antenna may be regarded as a series of elementary dipoles, and large antennas of other configurations may be considered as composed of many elementary dipoles. We will use

the elemental dipole antenna to illustrate the difference in coupling mechanisms in the far and near zones of an electromagnetic energy source.

FIELDS OF AN ELEMENTARY DIPOLE ANTENNA

An elementary dipole antenna oflength ~, which is short compared with a wavelength

(~« l) and the diameter is small compared with its length is shown in Fig. 2. It is energized

by a transmission line that does not radiate. Thus, for purposes of analytical simplicity, the

~T Q

TRANSMISSION 11 LINE

T Q

1 Figure 2. An elementary dipole antenna of length ~ which is short compared with a

wavelength, ~ < < l) and the diameter is small compared with its length.

elementary dipole may be considered as a thin conductor oflength, ~ carrying a uniform current,

I. It can be shown that the electric and magnetic fields from the dipole have only three components, Er, Ee, and Hcp. All other components are zero at all points [Jordan and Balrnain, 1968]. In the spherical coordinate system (Fig. 3), they are given by

(12)

(13)

Hcp = [Ue j(" t-~r) sin 8/41t] [j l3/r + lIr] (14)

where 13 is the propagation factor in the medium of intrinsic impedance 11.

Far Zone

At points far from the dipole antenna, r is large, terms involving lIr and lIr in (12-14) can be neglected in comparison with terms involving lIr. Thus, in the far zone, there are only two field components and are given by

7

(15)

H<» = j(II3~/4n:r) e j('" I·ar) sin e (16)

The wave impedance in the far zone is defined by the ratio ofEefH<», which is the same as the intrinsic impedance" of the medium. Also, Ee and H<» are in time phase and at right angles to each other. Thus, the electric and magnetic fields in the far zone of a dipole are related in the

same fashion as in a plane wave. Further, from the Poynting vector (P = E x H), one can find

the direction and time-average flow of energy per unit area given by

(17)

Clearly, energy flow in the far zone is real and is uniquely defined. It is outgoing in the radial direction. The energy is hence radiated, and the tenn radiation zone is synonymous with a far zone. In the far zone, the intensity of radiated energy decreases as lIr with increasing distance. As in a plane wave, the electric and magnetic fields are outgoing waves with plane wave fronts.

Near Zone

Inspection of the expressi~ns for E., Ee, and H<» shows that, at points close to the dipole antenna where r is small, the lIr and lITJ terms become predominant and equations (12) - (14) reduce to

(18)

(19)

(20)

Consequently, both components of the electric field are in time quadrature with the magnetic field. Therefore, the electric and magnetic fields in the near zone are related as in a standing wave. The maxima and minima of electric and magnetic fields do not occur at the same point in space. The ratio of electric to magnetic field strength varies from point to point, giving rise to widely divergent field impedances.

In addition, the tenns that vary asllTJ in the expression for E., and Ee correspond to the field of an oscillatory electrostatic dipole; these lITJ terms are referred to as electrostatic fields. The tenns that vary as l/r are the fields that would be obtained by a direct application of the Ampere law. Thus, the field represented by the lIr tenn is called the induction field and it becomes predominant at points close to the dipole antenna. It should be noted that at low frequencies, wavelengths are long, and the induction field may extend to very large distances from the source. The corresponding wavelengths at high frequencies are quite short, and the induction field may not exist at all.

Invoking the Poynting vector shows that the electrostatic and induction tenns represent energy that is stored in the field during one-quarter of a cycle and it is returned to the dipole antenna during the next one-quarter of a cycle without a net or average outward flow. In the near

8

zone, the energy exchange is largely reactive; only the lIr terms contribute to an average outward flow of energy. The energy transfer characteristics are illustrated in Fig. 3 where the arrows represent the direction of energy flow at successive instants of time [Michaelson and Lin, 1987].

FAR FIELD

I e-o· I I I

NEAR I FIELD' e

p

__ ....... ~ I

~-=;;-- ---~!-=------ ---_ - \ SHORT DIPOLE - ", ~,

RADIATED ENERGY

FLOW

REACTIVE'

ENERGY: FLOW I

I

Figure 3. The energy transfer characteristics of an elementary dipole antenna where the arrows represent the direction of energy flow at successive instants of time.

Power density in the near zone is not as uniquely defined as in the far zone, since the electric and magnetic fields and their ratios vary from point to point. Furthermore, the angular distribution actually depends on the distance from the antenna. It is necessary to individually arrive at a quantitative determination of the power density at all points.

The criterion of distance most commonly used to demarcate near or far zone is that the phase variation of the field from the antenna does not exceed AI16 [Silver, 1949]. This boundary occurs at a conservative distance of

(21)

where D is the largest dimension of the antenna aperture. In the far zone, the field strengths decrease as lIr, and only transverse field components appear. The near zone can be divided into two subregions: the radiative region and the reactive region. In the radiative region, the region closer than 2D2/A, the radiated power varies with distance from the antenna. The space surrounding the antenna in which the reactive component predominates is known as the reactive region. The precise extent of the regions varies for different antennas. For most antennas, the transition point between reactive and radiative regions occurs from 0.2 to 0.4 D2/A [Michaelson and Lin, 1987]. For the special case of an elementary dipole, the reactive component predominates to a distance of approximately A121t, where the radiative and reactive components are equal to each other. However, the outer limit is of the order of a few wavelength or less in most cases.

9

Quasi-Static Fields

At low frequencies where the wavelength of the electromagnetic radiation is at least an order of magnitude longer than the dimensions of the human body, its propagation behavior inside the body becomes quasi-static, and the electric and magnetic fields become decoupled [Lin, et al., 1973]. The induced field can be obtained by combining two independent electrostatic and magneto-static formulations of the electromagnetic field theory.

Under the conditions that the wavelength is long and the time variation is slow, the quasistatic electric and magnetic fields due to localized sources can be approximated using the static equivalent to equation (1) to (4) to yield,

(22)

Er = [Q~/(2ne~)] cos e (23)

(24)

for a localized electric source. In (23) and (24), Q is the total number of electric charges and Q~

is the electric dipole moment. Similarly, the quasi-static magnetic field due to a localized magnetic source or current element is given by

HCI> = [U 14nr] sin e (25)

where I is the current flowing in a short conductor of length Q. Since TIll\?> = jUlQ/e, equations (18) to (20) are the same as equations (23) to (25). It clearly indicates that the near fields of an elementary dipole are quasi-static. This observation extends to all antennas and radiating systems. Also, an incident plane electromagnetic wave couples to the human body in the same fashion as two separate electrostatic and magneto-static 1'i.elds. The wavelength of 50 or 60 Hz ELF fields is 5 to 6 million meters. Therefore, for all practical purposes, exposures to these ELF electric and magnetic fields occur always in the near zone inductive region. Their interactions are quasi-static in nature.

For example, an externally applied constant electric field gives rise to a constant induced electric field inside the body that has the same direction but is reduced in strength by a factor inversely proportional to the dielectric permittivity and is independent of body size. Similarly, the magnetically induced electric field inside the body is identical to the quasi-static solution of equation (1) and its magnitude is given by E = UlBr/2 = nfrlJH, where f= Ul/2n is the frequency, IJ is permeability, r is the radius, B is magnetic flux density, and H is the strength of the magnetic field component. Thus, the magnetic field produces an internal electric field that varies directly with distance away from the center and in proportion to frequency. Moreover, the magnetic field component of an incident plane wave under quasi-static conditions is given by the same expression. Thus, the induced electric field by a plane wave inside a human body is the result of a summation of its quasi-static electric and magnetic field components.

10

RELAXATION MECHANISM OF INTERACTION

Although the primary objective of this chapter is to discuss coupling mechanisms for ELF fields and RF radiation into biological systems, the inclusion of a well-known mechanism of interaction is advantageous from the perspective of energy coupling.

Polar molecules and other cellular components of biological materials can rotate in response to an applied sinusoidal electric field. The rotation is impeded by inertia and by viscous forces. Therefore, the orientation of polar molecules does not occur instantaneously, giving rise to a time-dependent behavior known as the relaxation process. Moreover, cells and tissue structures carry different electric charges. When subjected to a sudden electrical stimulation they require a finite duration of time for charges to accumulate at the interfaces and to equilibrate. The accumulation of charges at the interfaces continues until a condition of equilibrium is reestablished, thus, the relaxation. Many types of relaxation processes can take place in biological tissues, owing to polar molecules and membrane charges.

When a dipole distribution is uniform, the positi ve charges of one dipole cancel the effect of the negative charges from another adjacent dipole. Rut when the dipole distribution varies from point to point, this complete cancellation cannot occur. At an interface especially, the ends of the dipoles leave an uncanceled charge on the surface, which becomes an equivalent "bound' charge in the material. The relaxation process may therefore be examined by considering the response of bound charges to an applied electric field [Michaelson and Lin, 1986]. The dynamic force balance equation, for the model shown in Fig. 4, is

(25)'

where x is the displacement of a charged particle, E the applied electric field, cu. is the characteristic frequency of the elastic, spring-mass system, u is the particle collision frequency, and m and q are the particle mass and charge, respectively. The force exerted on the particle, mass multiplied by particle acceleration, on the left-hand side of the equation, results from an electric driving force qE, an elastic restoring force in proportion to displacement x with elastic constant denoted as mcu/, and a retarding damping force proportional to velocity dxldt with damping coefficient mu.

E

-x---.I

~ ~ DISPLACED POSITION IN EQUILIBRIUM RESPONSE TO E

POSITION

Figure 4. The response of a bound charge to an applied electric field.

11

After Fourier transformation and rearranging terms, (25) becomes

x(w) = [(q/m)E)/[w/ - w2+ jmu») (26)

Note that the equilibrium position for the charge (x = 0) represents local charge neutrality within the medium. When the charge is displaced from its equilibrium position, a dipole is established between the charge itself and the "hole" left behind and bound in the molecular and membrane structure. A dipole moment p is formed by the charge q times the displacement x. For a medium with volume-bound charge density, the total dipole moment per unit volume or polarization P is

P = pp = [p(q2/m)E)/[w/ - w2 + jmu). (27)

The electric flux density D may be expressed in terms of the electric field E and polarization P as

D= EoE+P. (28)

For isotropic media, the permittivity may be related to D by the expression D = EE. These relations together with (27), give an equation for the permittivity,

E(W) = Eo [1 +(w/)/(w/ - w2 + jwu») (29)

where (30)

and Eo is the vacuum or free-space permittivity. Clearly, E is a complex quantity and can be denoted by

E = E' - E" (31)

where E' and E" are the real and imaginary parts of the permittivity and can be obtained by equating the real and imaginary parts of (29) and (31).

The velocity of bound charge motion v = dxldt i.; obtained from (26), such that

v(w) = [(q/m)E)/[u - j(w.2 - cu 2)/w) (32)

The finite velocity of charge motion in the material media indicates that the particle cannot respond instantaneously to a suddenly applied electric field. This time-delay phenomenon gives rise to a frequency-dependent behavior of charge displacement leading to changes in permittivity with frequency or the relaxation mechanism of interaction of electromagnetic field and radiation with biological systems. It is note worthy that the same conclusions are reached by performing the inverse Fourier transforms of(29) and (32) and examining the phenomenon in a time domain. Note that the dependence of permittivity on source and characteristic frequencies w, wP' and w. suggests that the charge displacement (26) and motion (32) could also be resonant in nature.

12

INDUCED FIELDS AND ENERGY ABSORPTION

Clearly, regardless of the mechanism of interaction, fields must be coupled into the system and energy must be absorbed or deposited in the biological systems in order for the system to respond in some manner. Thus, to realize any biological response, the electric, magnetic or electromagnetic field that is effective in exerting its influence must be quantified and correlated with the observed phenomenon. The dosimetric quantities of import include an incident field, induced field, specific absorption rate (SAR), and specific absorption (SA) in biological systems or tissue models. The metric SAR is defined as the time derivative of the incremental energy absorbed by (or dissipated in) an incremental mass contained in a volume of a given density [NCRP, 1981]. Specific absorption (SA in J/kg) is the total amount of energy deposited or absorbed and is given by the integral of SAR over a finite period of time. Information on SA and SAR is of interest because it may serve as an index for extrapolation of experimental results from animal to animal and from animal to human exposure. It is also useful in analyzing relationships among various observed biological effects in different experimental models and subjects.

A determination of the induced field would be preferred because (1) it relates the field to specific responses of the body, (2) it facilitates understanding of biological phenomena, and (3) it is independent of mechanisms of interaction. Once the induced field is known, quantities such as SAR in W /kg can be derived from it by a simple conversion formula. For example, from an induced electric field E in Vim,

(33)

where a is the bulk electrical conductivity and Pm is the mass density (kg/m3) of tissue, respectively. However, at present, a small, isotropic, implantable electric field probe has yet to be developed with sufficient sensitivity for practical use. Consequently, a common practice in experimental dosimetry relies on the use of temperature elevation produced under a short-duration «30 s), high-intensity exposure condition. (The short duration is insufficient for significant convective or conductive heat contribution to tissue temperature rises.) In this case, the time rate of initial rises in temperature (slope) can be related to SAR through a secondary procedure, such that

SAR= cATIAt (34)

where AT is the temperature increment eC), c is the specific heat capacity of tissue (J/kg_0C), and At is the duration (s) over which AT is measured. It is important to distinguish the use of SAR and its derivation from temperature measurement. The quantity ofSAR is merely a metric for energy deposition or absorption and it should not be construed to imply any mechanism of interaction, thermal or otherwise. However, it is a quantity that pertains to a macroscopic phenomenon by virtue of the use of bulk electrical conductivity in its derivation (33).

The quantity SAR was officially designated by NCRP as a metric for nonionizing radiation dosimetry. It is defined as the time derivative of the incremental energy absorbed by (or dissipated in) an incremental mass contained in a volume of a given density [NCRP, 1981]. The acceptance of SAR as a meaningful dosimetric measure enabled ANSI to base its 1982 revision of ANSI C95.1 standards on dosimetric concepts. This was the first time when the ANSI

13

C95.1 safety standards incorporated into it the frequency-dependent absorption ofRF radiation by the human body.

It is important to emphasize the use of bulk electrical conductivity, the specific heat capacity, and the mass density (kg/m3) of tissue in the derivation of SAR from electric field strength and temperature elevation. Their use in the definition meant that a volume of tis sue mass must be selected over which SAR is determined. Until the recent suggestion of 10 grams, one gram of tis sue in the form of a cubic volume was specified. It is obvious that the numerical value ofSAR would be the same regardless what volume is chosen if the power deposition is uniform. A difficulty arises when the absorption is nonuniform or tissue with differing properties are within the same volume. In theory, a 10-g averaging mass could underestimate SARs of nonuniform fields by up to a factor of 10 compared with a I-g averaging mass. Measurements made in a homogeneous brain liquid phantom and using five commercially available telephone models showed a difference of two between the I-g and 10-g says [Kuster and Balzano, 1997]. Likewise, computer modeling based on head and neck anatomy has shown I-g SARs twice as high as values computed for 10 g [Gandhi et al., 1999]. Tllus, a 10-g SARmetrlc would tend to be more conservative.

FREQUENCY DEPENDENCE OF ABSORPTION IN BIOLOGICAL BODIES

The coupling of electromagnetic energy into biological tissues is governed by the frequency and configuration of the source, and the geometry and composition of the exposed body. Although coupling mechanisms at ELF and RF frequencies are different, a brief account is presented for plane wave coupling to the whole body as a function of frequency. A plane wave is defined by a uniform wave front in which the electric and magnetic field components are related through a constant impedance, i.e., the intrinsic impedance of the medium. Moreover, the electric and magnetic fields are orthogonal in space and both are perpendicular to the direction of wave propagation.

For all practical purposes, exposure to ELF electric and magnetic fields take place always in the near zone inductive region regardless of distance from the source of such fields. An equivalent plane electromagnetic wave at ELF would be two separate and independent quasistatic electric and magnetic fields with a constant field impedance equal to that of free space, '10 = 1201t ohms. As previously indicated, at ELF frequencies, the electric and magnetic field components are decoupled inside a biological body.

As a first order approximation, the frequency dependence of SAR for a homogeneous

spherical model of a human adult weighing 70 kg in a plane wave field is shown in Fig. 5. The radius of the muscle sphere is 25.6 cm and an incident equivalent power density of 10 W/m2 is assumed. In particular, the maximum SAR, average SAR and the average power deposition per unit surface area are shown for the frequency range between 1.0 and 10,000 MHz. Note that SAR and power deposition both vary greatly for different frequencies. For frequencies above 100

to 200 MHz, average power deposition and SAR are nearly constant at 2 x 10-3 W /kg per W/m2

of incident power density at 200MHz as indicated by the dashed line. In the frequency range of 20 to 200 MHz, the average SAR and power deposition decreases directly in proportion to frequency. For frequencies below 20 MHz, the average SAR and power deposition decrease as the square of the frequency. Though not shown in this figure, it has been demonstrated that the

14

~I~~~~~~---r------~------' 25.6 em. RADUS

SARmax W/kg

10°

! Q. -_. Pave mW/cm 2

C KJ1 CD "C () .E '" E ..e ~

r-_-t __ .:S::AR ave W/kg

E .... CD Co

~ II: c: .Q e-o t/) .c « g '0 CD Co en

~~I~~~~~~~~~~~~~~ 10 100 1000 10000

Frequency (MHz)

Figure 5. The frequency dependence of SAR for a homogeneous spherical model of a human adult weighing 70 kg in a plane wave field. [Redrawn from Lin et al. 1973]

15

variation with square of the frequency extends monotonically down to 1 Hz [Lin et al. 1973]. The sharp falloff in SAR and power deposition as the frequency is reduced indicates that the coupling of electromagnetic energy to human body is much reduced at lower frequencies including ELF as compared to the RF range. The deviation of peak SAR at higher frequencies reflects the fact that at these frequencies the same amount of incident power is deposited in decreasingly thinner superficial layers of the model.

Table 1. Relative dielectric permittivity for a homogeneous brain model

Frequency (Hz)

10 102

103

104

105

106

10' 108

9.15 x 108

2.45 x 109

1010

£1

6,570,000 525,000

85,400 39,400 13,100 953 105 43.0

34.4

30.9

24.2

123,000,000 13,200,000 1,480,000 154,000 22,500 4,730 739 95.8

15.5

10.7

15.2

The average power deposition and SAR in a human head exposed to a plane wave field can be model using a homogeneous brain sphere. Fig. 6 illustrates the dependence on frequency for a 7-cm radius brain sphere with the dielectric permittivity given in Table 1. Note that energy absorption and power deposition are about an order of magnitude small than the corresponding values for the whole body model. The general behavior of frequency dependence is similar to that of the whole body model. Average absorptions agai.n reach a maximum around 200 MHz and remain fairly constant or decrease slightly with increasing frequency. This is consistent with some of experimental and anatomically based numerical data indicating that the highest part-body average SAR for the head and neck region occurs at 200 MHz [Gandhi et al. 1992]. But the absorptions falloff as f2 for frequencies below 100 MHz. At ELF, the coupling would be minuscule from a plane wave source. The electric field component in a plane wave is 1201t times greater than the magnetic field component. The high tissue dielectric permittivity would limit severely the ability of a low frequency electric field applied through air from penetrating into biological tissues. This is a succinct demonstration of the salient distinction between ELF and RF field coupling to biological systems. ELF fields are quasi-static and the electric and magnetic field components are decoupled in their interaction with biological systems. Moreover, their coupling mechanism is inductive in character. Their behavior is such that to produce the

16

10' ... ~ 10° a. .... c: CI) 10"9

"0 10"'

'0 .E: 10-2 -0

"'E lO"n 0

~ 10-12 E

Pave (mW/cm2) 10"3 SARmax (W/kg)

SARave (W/kg) 10'"

... CI) Q. 10"5

.m as 0:: 10-6

c: 0 :;:; Q.

10"7 ... 0 !/) .0 10'8 <C 0

I;::: '0 CI) Q.

CI)

Frequency (Hz)

Figure 6. The dependence relative absorption on frequency for a 7-cm radius brain sphere.

17

same level of induced electric field inside the body or model by an ELF magnetic field, the applied ELF electric field must be 1201t times greater than the electric field of a plane wave.

ELF ELECTRIC AND MAGNETIC FIELD COUPLING

Aside from SAR and power deposition, in many applications including exposure assessment, the quantities of prime interest are the SAR, induced field and current distributions

within the body. The wavelength of ELF fields is about 5 x 106 meters. For all practical

purposes, exposures to ELF electric and magnetic fields occur always in the near zone inductive region. The induced electric and magnetic fields inside a human body are therefore quasi-static in character. Moreover, at ELF frequencies, the electric and magnetic field components are decoupled inside a biological body. Indeed, these phenomena arise whenever a biological body or model is small compared to a wavelength. Also, at ELF frequencies, the bulk electrical conductivity is of the order of I S/m and the relative diele("tric permittivity is about 106• The ratio

a/WE »1 (35)

Thus, the conduction current is much greater than displacement current, and biological materials may be considered as conducting media.

Electric Field

A reasonable estimation of the electric field can be derived from

E =-VV (36)

where the potential V is the solution to Laplace's equation,

v • aVV=O (37)

For a uniform or constant electric field Eo, polarL:ed in the x direction, the induced field inside a spherical model with radius r and dielectric permittivity E is given by

(38)

Therefore, an externally applied constant electric field gives rise to a constant induced electric field inside the body which has the same direction but is reduced in strength by a factor inversely proportional to the dielectric permittivity of tissue (Fig. 7). A surface polarization is set up which generates a uniform internal electric field. While equation (38) suggests that the induced field is independent of body size it varies from tissue to tissue as a function of dielectric permittivity.

18

x x

"+-+r1 ...... z

electric component magnetic component

Figure 7. The coupling of low frequency.electric and magnetic fields into a homogeneous spherical model of biological tissue.

JOE, I

Figure 8. ELF electric field distribution measured at the body surface of a human body subject standing under an electric power transmission line. Eo is the electric field strength in air.[Redrawn from Shimizu et al. 1988]

19

At ELF frequencies, the high tissue dielectric permittivity and conductivity and the behaviors of electric field at curved boundaries combine \0 render some very unique phenomena. They conspire to weaken the low frequency electric fields applied through air by about 10-6 upon penetration into biological tissues. The same boundary condition distorts the applied uniform electric field in the immediate vicinity of the biological body in such a manner that it becomes oriented everywhere perpendicular to the surface of the body. Moreover, the surface electric field shows considerable enhancement at each sharp curvature of the body. This phenomenon is illustrated in Fig. 8 using the distribution of electric field at the surface of a human subject standing in an ELF field produced under a high voltage transmission line. The current densities through selected axial cross sections of a grounded human, swine, or murine body exposed to vertical, 60 Hz, 10 kV 1m electric fields are shown in Fig. 9. For example, in the case of humans,

180kVIm

Figure 9. The current density through selected axial cross sections of a human, swine, or murine body exposed to vertical, 60 Hz, 10 kV/m electric field. Relative body sizes are not to scale. Average axial current densities are estimated values over each cross section. Current densities perpendicular to surface of the body shown for man and pig are from calculations. Surface electric fields are measured values for man and pig and that for rats are estimated. [Redrawn from Kaune and Phillips, 1980]

a current is induced to flow in the vertical direction Current density is the highest at the narrower cross sections of the leg or neck which are estimated to be 0.02 or 0.006A1m2, respectively. It is noteworthy that the surface electric field is proportional to the height of the body in the field. While current and field distributions within different body components may offer greater quantitative details, these averaged values clearly demonstrate the significance of body cross sections in determining the current distribution. Also, tis!'ue composition of specific body parts contributes to the induced field and current density via their electrical conductivity.

20

Magnetic Field

It is emphasized that since the magnetic penneability of biological materials is approximately the same as free space, the magnetic field inside a body is equal to that applied externally. However, a vertically directed unifonn magnetic field induces an electric field inside the body which is identical to the quasi-static solution of equation (1). Its magnitude is given by

(39)

where f is the frequency, J.I. is permeability, r is radial distance away from the center, and Ho is the strength of the magnetic field component. Thus, the magnetic field produces an internal electric field that varies directly with distance away from the center and in proportion to frequency and the applied unifonn magnetic fields. The induced current density is

J = oE = (owrJ.l./2)Ho <P (40)

Thus, the magnetically induced electric field encircles the magnetic axis and produces an eddy current whose magnitude increases with distance from the center of the body (Fig.7).

It is cautioned that when applying (39) and (40) t,) estimate induced fields inside a body (animals, humans, or tissue preparations), any significant deviations from homogeneity or circular cylindrical symmetry must be taken into account. Equations (39) and (40) should be applied to each region inside the body with a different conductivity which behaves as a unit with its own body center and radius or an equivalent radius. However, due to opposing field orientations and current paths the highest field and current densities tend to occur with the large dimensions associated with outer layers of a body or tissue preparation so long as the conductivities are not grossly different and the regions are not separated by nonconducting materials.

The circulating eddy currents induced inside a prolate spheroid and a homogeneous model of the human body immersed in a unifonn horizontal or vertical magnetic field are shown in Fig 10. The induced electric fields are polarization dependent in that they are always in a plane perpendicular to the magnetic field [Durney et al. 1975]. The eddy current loop is the largest when the dimension of the homogeneous region is the largest.

Electric and Magnetic Fields

It is noteworthy that, when both are present, tht relative significance of electrically or magnetically induced absorption in humans and animals is a function of body size and field impedance, 11 (ratio of applied EjH.,), such that

E.!EJ, = (2e.,loJ.l.r)(EjH.,) = (2e.,loJ.l.r)11 (41)

Accordingly, the magnetically induced field in a human being and a mouse is 601t greater than the electric case for a ratio of applied EjH., = 1201t ohms. The electric and magnetic field induced absorption would be equally significant if the applied electric strength is 601t times stronger than the applied magnetic field. The same is true if the size of the body is 1I601t of a

21

H

Horizontal HField

~ Spheroidal Model

2m

Vertical H Field

Figure 10. The circulating eddy currents induced inside a prolate spheroid and a homogeneous model of the human body immersed in a uniform horizontal or vertical magnetic field.

meter. Thus, for an isolated cell or a small aggregate of cells, the electrically induced field would be the predominant factor when the ratio of applied EJHo = 1201t ohms, the intrinsic impedance of the medium. This size dependence illustrates the need for scaling in extrapolating results from animals to humans or from cell preparations to whole body systems. Substantive adjustment is required when a comparable induced field inside different animals are to be obtained by applying external fields.

A Summary

The coupling and distribution characteristics of ELF and other low frequency electric and magnetic fields in biological tissue have been described. lbe results apply to frequencies where the wavelength is long or the largest dimension of the body is small compared with a wavelength. The results indicated that

Exposures to ELF and other low frequency electric and magnetic fields occur in the near zone inductive region of the source.

• Induced electric and magnetic fields inside animals and humans are quasi-static in nature.

•

•

22

Magnetic field inside a body is equal to that applied externally.

Electric and magnetic fields are decoupled inside a biological body.

Electric fields applied through air is weakened by about 10-6 upon penetration into biological tissues.

Induced electric field in the immediate vicinity of the biological body is perpendicular to the surface of the body.

•

•

•

•

•

•

•

Electric field is enhanced near the surface at sharp curvatures.

Magnetically induced electric field encircles the magnetic axis and produces an eddy current whose magnitude increases with distance from the center of the body.

Eddy current calculation must be applied to each region inside the body with a different conductivity which behaves as a unit with its own body center and radius or an equivalent radius.

Highest induced electric field and eddy current density occur with the large dimensions associated with outer layers of a body with finite conductivity.

Relative significance of electrically or magnetically induced absorption in humans and animals is a function of body size and field impedance.

Conduction current is much greater than displacement current in biological materials.

Biological materials may be considered as conducting media.

Biological bodies or models are small compared to a wavelength at ELF and other low frequencies.

COUPLING MECHANISMS FOR RF RADIATION

The coupling of incident RF radiation into biological tissues is influenced by the geometry and composition of the exposed body and the frequency and configuration of the source. In addition, the character and strength of the incident field would differ according to distance from the source and the specific source type, i.e., handheld, land mobile, base or broadcast station. As an example, the incident field strength ranges from 3-10 Vim (24 to 265 mW/m2) at typical far zone distances from a wireless telephone base station [Lin, 1997].

Furthermore, as shown in Fig. 5, the maximum SAR, average SAR and the average power deposition per unit surface area are highly frequency dependent for an adult human model weighing 70 kg. Specifically, for frequencies up to 20 MHz, the average SAR and power deposition increase as the square of the frequency. In the range of 20 to 200 MHz, the average SAR and power deposition increase directly in proportion to frequency. For frequencies above

200 MHz, average power deposition and SAR are nearly constant at 2 x 10-3 W /kg per W 1m 2 of

incident power density. An important distinction in assessing the coupling ofRF energy into biological systems

is the determination whether the exposure is taking place in the near or far zone of a given source. The 2D2/l distance (21) is approximately 6 cm for al0-cm RF antenna operating at 900 MHz in free space. Clearly, both near-zone reactive and far-zone radiative interactions are encountered in the vicinity of wireless RF telecommunication systems. Note that the radiation efficiency of an antenna depends on the physical dimension (0) and the operating frequency or wavelength (l). A 10-cm antenna would radiate a significant amount of energy since D/l (= 0.33) > 0.1 at 900 MHz,

23

As equations (15) to (17) indicate, in the far zone of an antenna, the RF energy radiated energy propagates as a plane wave. The interaction of RF radiation with biological systems is independent of the source configuration in the far zone. The electric and magnetic fields are uniquely defined and are related through a constant factor, the intrinsic impedance of the medium. Therefore, the determination of electric field behavior is sufficient to characterize the interaction. Accordingly, this discussion will begin with plane wave RF field interaction.

Reflection and Transmission

At boundaries separating regions of different biological materials, RF energy is reflected or transmitted. For a plane wave impinging normally from a medium of intrinsic impedance 11\, on a medium of intrinsic impedance 112' the reflection coefficient is given by

(42)

The transmission coefflcient provides a measure ofRF energy coupling and is given by T = 1 + R. The fraction of incident power reflected by the discontinuity is R2 and the transmitted fraction is (1 - R2). For very similar tissues where 11\ is equal to 112' there is no reflection and the transmission is maximum. As the transmitted wave propagates in tissue medium, RF energy is extracted from the wave and deposited in the medium, resulting in a progressive reduction of power density of the wave as it advances in the tissue. This reduction is quantified by the depth of penetration or skin depth, which is the distance througi l which the power density decreases by a factor of e-2. Note that the rate of energy deposition is equal to SAR. Table 2 gives the calculated skin depth and transmission coefficient from air, using typical dielectric permittivities, for tissues of high water content such muscle and most organs,. and tissues of low water content which include bone and fat.

Table 2. Propagation characteristics of RF plane waves in low- and high-water (H20) content biological tissues at 37°C.

Frequency Dielectric Conductivity Skin Depth Transmission (MHz) Constant (S/m) (cm) Coefficient

H2O High Low High Low High Low High Low

27 113 20.0 0.61 0.03 14.3 77 0.14 0.56 40 97 14.6 0.69 0.03 11.2 58.8 0.17 0.62 100 72 7.5 0.89 0.05 27 106 0.22 0.74 433 53 5.6 1.43 0.08 3.6 183 0.36 0.82 915 51 5.6 1.60 0.10 2.5 12.8 0.40 0.83 2450 47 5.5 2.21 0.16 1.7 8.1 0.43 0.84 5800 43 5.1 4.73 0.26 0.8 4.7 0.44 0.85 10000 40 4.5 10.3 0.44 0.3 2.6 0.45 0.87

24

Clearly, the coupling of RF energy from air into planar tissue is higher for low water content tissue than that of high water content tissue. It is higher for higher RF frequencies than for lower ones and ranges from about 15 to 80%. The data given in Table 2 show that the penetration depth for low water content tissues such as tone and fat is nearly five times greater than for higher water content tissues. And it is also frequency dependent. The transmission coefficient for tissue-tissue interfaces generally is larger than air-tissue interfaces and the reflection coefficient is just the opposite. The reflection coefficient varies from a low of five for muscle-blood to a high of about 0.50 for a bone-muscle interface. [Lin and Gandhi, 1996].

In layered tissue structure with different dielectric permittivity, the coupling behavior can be very complex. Multiple reflection could occur between tissue interfaces. The transmitted wave will combine with the reflected wave to form standing waves in each tissue layer. The peaks of the standing wave oscillations can result in higher SAR to yield greater coupling of RF energy into the tissue layer. The standing wave phenomenon becomes especially pronounced if the thickness of each layer is less than the skin depth for that tissue and it is approximately one-half wavelength or longer at the RF frequency. This dependence of SAR oscillation peaks on layer thickness is a manifestation of layer resonance. Fig. 11 illustrates the dependence of peak SAR on thickness of the fat layer in a planar tissue model of skin-fat-muscle layers at different frequencies. The skin is assumed to be 2 mm thick. The incident power density is 10 W/m2• It

8' ::.:: ~ W

~ z o ii: 0::: o W co « (.)

3.0

2.0

1.0

ii: 0.1 (3 w 0-w ::.:: 0.05

L5 0-

915 MHz

, , I

I I

I

2450 MHz

SKIN----- MUSCLE--

-----------------~~..!:'!.. ---27.1 MHz

0.01 --.J o 0.5 1.0 1.5 2.0 2.5 3.0

FAT THICKNESS (em)

Figure 11. Dependence of peak SAR on thickness of the fat layer in a planar tissue model of skin-fat-muscle layers at different frequencies. The skin is assumed to be 2 mm thick. The incident power density is lOW 1m 2•

25

is noteworthy that for fat layers up to 3 cm thick, the peak SAR is higher in the skin layer for each of the RF frequencies shown. The distribution ofSAR for RF energy propagating through a subcutaneous fat layer into muscle is shown in Fig. 12. The values are normalized to the SAR

in muscle at the fat-muscle interface. These SARs ""ill remain the same for smaller fat thicknesses, i.e., the distribution of relative SAR corresponding to a 1 cm thick fat layer would be the same as shown in the figure for fat and muscle. Lower SARs occur in fat compared to that in muscle. The fat-muscle interface produces a standing wave resulting in a hot spot in the fat layer, one-quarter wavelength from the muscle interface. The standing wave oscillations become bigger in fat and the coupling into muscle becomes smaller as the frequency increases.

Coupling to Bodies with Curvature

A plane wave RF field in the far zone can overcome skin depth losses and produce enhanced coupling at greater depth in bodies with curved surfaces. Typically, if the largest dimension of the body is comparable to the wavelength and beam of the impinging RF radiation,

~ II: c::: o ~ o 1l <{ u ;: .~

Co (/J

.~ 1ii (j) II:

1.0

0.8

0.6

0.4

0.2

FAT BONE

-2 -I 0

MUSCLE

2 3 4 5

Z (em)

27 MHz 433 MHz 915 MHz

2450 MHz

6 7

Figure 12. The distribution of SAR for RF energy propagating through a subcutaneous fat layer into muscle. Values are normalized to SAR in muscle at the fat-muscle interface.

26

SAR and its distribution will be influenced by the whole body or body part, curvature of its surface, and the tissue composition. In particular, the ratio of geometric variables and wavelength dictates the characteristics ofRF energy coupling. An example of this phenomenon is given in Fig. 13 where the relative absorption coefficients for a 7 -cm radius homogeneous sphere of brain

2.0

~1.8 w ;::;1.6 lL.

l!:i1.4 o u zl.2 o ;::1.0 a... a::: ~.8 !D a:

w· 6 >

~.4 -1

~.2

2 5

/

j /

/ ~

V 2

O ka 10 2

1/ BRAIN J l.- t---

II 1/ ~ '-. J f'-..-MUSCLE r--

5 103 2 FREQUENCY (MHZ)

5

r-- t--r-1-1-

5

Figure 13. The relative absorption coefficients for a 7-cm radius homogeneous sphere of brain and muscle equivalent materials as a function of frequency.

and muscle equivalent materials as a function of frequency are shown. It can be seen that for frequencies within the range of 400 to 1500 MHz and 550 to 800 MHz for brain and muscle, respectively, the incident RF plane wave is resonantly absorbed by tissue spheres. The average absorption exceeds by as much as 1.2 that impinging in the shadow cross section [Lin, 1986]. At these frequencies, coupling is enhanced by the presence of resonant standing waves inside the tissue model. The SAR distributions inside a brain sphcr. > would show maxima or hot spots near the center of the head model. As the frequency increases, the SAR peak shifts gradually to the front surface and falls off slowly in magnitude, indicating that the details of the body curvature are of diminishing significance.

The coupling of plane RF radiation into more complex models of the head structure where a spherical core of brain is surrounded by five concentric shells of other tissues (See Table 3) is illustrated by Fig. 14. The SAR distribution is normalized to the maximum along the z direction. The incident plane wave power density is 10 W/m2, propagating in the positive z direction. It can be seen that the effect of skin, fat, skull, dura, and cerebral spinal fluid on SAR

27

Figure 14.

28

Table 3. Structure and composition of a six-layer model of the human head 10 cm in radius

Tissue

Brain CSF Dura Skull Fat Skin

Layer thickness mm

Dielectric Permittivity 900 MHz 2400 MHz

I

69.8 11.0 8.0 7.0 2.7 1.5

34.4 - j15.5 80.9 - j14.0 51.4 - j25.1 5.56 - jO.86 5.56 - jO.86 51.4 - j25.1

FRECl = 918 MHZ Pi = 1 mW/CM2

BRAIN SPHERE . MAX ABSORB = 1.589 W/KG LINEARLY POLAR AVG ABSORB = LlI8 W/KG

/'- X I~ /

Y

10- CM -10.00 -6.00 -2.00 2.00 6.00 10.00

FRECl = 2450 MHZ Pi = 1 mW/CM2

BRAIN SPHERE MAX ABSORB = 1.665 WI KG Llf\EARL Y POLAR AVG ABSORB = 0.062 W/KG

I

"" X//

N!". L/1Z

CM -10.00 -6.00 -2.00 2.00 6.00 10.00

32.8 - j15.4 77.0 - j13.9 47.5 - jl1.4

5.0 - jO.86 5.0 - jO.86

47.5-jl1.4

The coupling of plane RF radiation into models of the head structure where a spherical core of brain is surrounded by five concentric shells of other tissues. The SAR distribution is normalized to the maximum along the z direction. The incident plane wave power density is 10 W/m2 propagating in the positive z direction.

distribution is increased SAR in the skin. Absorptions in fat and skull are the lowest among the tissue layers [Lin and Gandhi, 1996]. Moreover, the peak and average SARs may be several times greater than for homogeneous models. The enhancement is apparently due to resonant coupling of plane wave RF into the brain sphere by the outer tissue layers. Although SAR in the shadow region is lower compared with the front surface, it is still quite substantial. In addition, SARs at the top and bottom or left and right sides of the model have equal magnitudes. It is noted that RF coupling is weaker and the skin depth is shorted for non plane wave fields, especially for incident radiation of limited beam width.

Orientation and Polarization Dependence

For elongated bodies such as those shown in Fig. 10, where the height-to-width ratio is large, the coupling ofRF energy is influenced by the orientation of the electric field vector with respect to the body (polarization). The three principal polarizations of the impinging plane wave to be distinguished are: E-polarization in which the electric vector is parallel to the major axis of the body, H-polarization in which the magnetic vector is parallel to the major axis of the body, and K-polarization in which both the electric and magnetic vectors are parallel to the major axis of the body. The frequency at which resonant absorption occurs is a function of both polarization and the exposed subject. In general, the shorter the subject, the higher the resonance frequency and vice versa. And E-polarization couples RF energy most efficiently to the body in a plane wave field for frequencies up to and slightly above the resonance region.

For RF frequencies well below resonance such that the ratio oflong body dimension (L) to free space wavelength (A) is less than 0.2, the average SAR is characterized by a f dependence. SAR goes through the resonance region in which 0.2 < LlA < 1.0. Specifically, the SAR rapidly increases to a maximum near LlA = 0.4 and then falls off as lIf. At frequencies for which LlA > 1.0, the whole body absorption decreases slightly but approaches asymptotically the geometric limit of about one half of the incident power, i.e., 1- power reflection coefficient. The resonances are not nearly as well defined for H-polarization as for E-polarization. The average SAR for H-polarization gradually reaches a plateau throughout the RF spectrum [Lin and Gandhi, 1996].

RF Coupling in the Near Zone

The discussion ofRF energy coupling has been confined to the far zone or the radiation zone of the antenna. The electromagnetic fields are outgoing waves with plane wave fronts. However, the accelerated use ofRF radiation for wireless communication over the past few years has focused considerable attention on the amount ofRF energy coupled into human bodies and on any potential biological effects. Plane wave analyses are applicable to base-station antennas found across many urban and suburban landscapes and :n residential and office environments. However, the antenna of a cellular or mobile telephone is typically located next to the user's head, thus creating an exposure situation in the near zone of the RF antenna. As mentioned already, the outer limit of the near zone is on the order of a few wavelengths or less in most antenna systems. Even for an elementary dipole, the reactive near zone distance is approximately AI21t, which is about 5 cm in air at 900 MHz.

In contrast to plane waves in the far zone, near-zone RF electric and magnetic fields are in time quadrature (See equations 18-20). The wave impedance is no longer the same as the

29

intrinsic impedance and varies from point to point in the near zone. The maxima of electric and magnetic fields also do not occur at the same location ill space. Since these are precisely the characteristics of a standing wave, thus, RF radiation in the near zone behaves like a standingwave field. RF energy will be transferred back and forth between the radiating antenna and the body.

While these features are reminiscent of a human being exposed to ELF fields from a power transmission line, there are some important differences. In the ELF case, the exposure takes place in an environment where the electric and magnetic fields are uniform and very large in extent compared to the size of a typical human body. While ELF energy may change slowly in time, they either are stored in or transferred between the electric and magnetic fields in the immediate vicinity of the source, but not radiated. However, a dipole-type wireless RF antenna is small compared with the size of a human head. The beam width in the near zone is also smaller than the head. As shown in Fig 3 the field is diverging in the near zone. In addition to me reactive induction field, there is a radiative component that is outgoing which is proportional to the product between the electric and magnetic field components given by (19) and (20) for a small dipole antenna. Since RF currents produce magnetic fields, and time-varying magnetic fields generate electric fields in tissue, RF coupling in biological tissues can be properly expressed in terms of induced fields. However, the generally used dosimetric quantity SAR is equally applicable as a derived quantity for RF in both the near and far zone.

Coupling from Handheld Cellular Telephones

Cellular telephones and personal telecommunication devices are designed to operate in close proximity of the user and typically are located next to the user's head. Aside from the intended purpose of radiation into the environment, RF energy from these devices is coupled to the head, neck, or hand of the user found in the near zone of the radiating system. A variety of experimental and computational methods have been used to quantify induced fields and SARs, and to help in assessing the health and safety risk of these telecommunication devices [Lin and Gandhi, 1996; Kuster et al. 1997]. This discussion will summarize SAR distributions inside homogeneous and inhomogeneous phantom head models. A particular interest will be the coupling mechanism of RF energy from hand-held cellular telephones and other personal communication systems.

Investigations to measure SARdistributions in canonical (spherical) or anatomically shaped models of the human head have employed isotropic electric probes in skulls filled with brainequivalent, liquid dielectric phantom materials [Cleveland and Athey, 1989; Balzano et al., 1978, 1995; Kuster and Balzano, 1992; Schmid et al., 1996; Gandhi et al., 1999]. These efforts are complicated by the variability of electric and magnetic fields in the near zone of the antenna, and also by the wide array of possible device positions and tilt angles of the antenna during normal operation. Thus, an intended use position and an a'lgle of 30° are often adopted in these investigations for a given device. Nevertheless, a survey of the reported measurements, in homogeneous models of head phantoms, showed that the measured SARs would vary depending on the specific antenna configuration and placement of the antenna next to the head (Table 4). Other contributing factors include difficulties associated with performing accurate and reliable SAR measurements when the source is in close proximity to the head.

30

Table 4. Some measured I-g SARs in head phantoms exposed to wireless communication devices (600 mW output power)

Max.SAR Frequency (Brain) Distance Antenna Author (MHz) (W/kg) (cm) Type Date

815 1.0 1-2 114 A Cleveland & Athey 855 2.1 1-2 YlA [1989] 835 0.63 1.0 114 A Anderson & Joyner

0.44-0.83 1.0 YlA [1995] 835 1.8 (Ear) 2.5 114 A Balzano et al.

[1995] 835 3.21- Gandhi et al.

5.41 [1999] 900 3.6 2.5 ~A Kuster & Balzano

[1992]

Computational procedures applied, to date, have modeled the cellular transceivers mostly as a metal box with an antenna. The advantage of numerical SAR computation is a more realistic description of the telephone and its position relative to the user, and the ability to vary these parameters to examine the dosimetric interaction ofRF radiation from cellular telephone operation with the human body such as the head, hand, and the entire body [Lin and Gandhi, 1996; Paulsen, 1997]. Typically, a magnetic resonance imaging (MRI) based model of a human adult is used for SAR calculations. The computerized MRI head models have a spatial resolution of 1-3 mm. Frequency-specific tissue permittivities such as those given in Table 5 are used to represent each tissue type in the inhomogeneous model of the head.

Common or intended head-phone positions have been used to compute SAR distributions in the head. Table 6 presents a list of computed SAR for cellular telephones operating between 800-1900 MHz. The antennas simulated include Al4, 3/8 A, Al2, and internal structures. Note that these results do not exhibit any consistent effect of antenna type or length on induced SAR. However, the peak SAR tends to be lower for longer antennas compared to shorter (Al4) antennas. Questions have been raised concerning the inter-comparability of computational results using different models of the human head and the mobile telephone transceiver.

There are also computer calculations ofSAR using.canonical models of the human head and idealized antennas [Dimbylow, 1993; Chen and Wang, 1994]. Again, depending on the model selected for computation, the induced maximum SAR varies considerably from model to model. More important, they can be very different from each other and from the data given in Tables 6. For example, an SAR of2.63 W/kg inside a block type phantom head was obtained for 600 mW of power radiating from an idealized 835 MHz dipole [Chen and Wang, 1994]. These results suggest the need for standardization in various modeling efforts to simulate mobile telephone exposure, and the difficulty encountered in dosimetric assessments. Accordingly, it may be concluded the anatomical configuration of the head and tissue inhomogeneity can all influence the

31

Table S. Permittivity (conductivity and dielectric constant) ofbiologicaI tissues for the radio frequency used by mobile telephones (800-900 MHz)

Tissue Type Dielectric Conductivity Constant (S/m)

Air 1.0 0.0 Fat 11 0.17 Bone 21 0.33 Cartilage 37 0.8 Skin 35 0.6 Muscle 50 1.08 Lung 12 0.24 Brain 41 0.86 Cerebral spinal fluid 78 1.97 Blood 55 1.86 Eye 67 1.97 Vitreous humor 67 1.68 Cornea/sclera 51 1.13 Lens 45 0.75

Data from [Gabriel et aI., 1996]

maximum vaIue and distribution of SAR in the head of a mobile telephone user. However, available information from literature seems to indicate that the integrated SAR in the head is similar for a homogeneous or inhomogeneous model.

Fig. IS gives a map ofSAR computed for a '),)4 monopole antenna oriented vertically with respect to the head. The SAR ranges from 0.1 to 6.0 W/kg with the maximum appearing at the appending point of the ear to the side of head. It can be seen that most of the power deposition is on the side of the head nearest to the radiating structure of the cellular telephone [See Gandhi et aI., 1996]. The SAR is considerably lower elsewhere in the head.

It is significantto observe from the energy coupling perspective that 40-50 % of the radiated RF power is absorbed by the human body [Tinniswood et aI. 1998; Gandhi et al. 1999]. The bulk of power deposition is on the side of the head nearest to the radiating structure of the cellular telephone. The SAR distribution follows an exponential trend away from an antenna side and it is considerably lower elsewhere in the head. The maximum SAR and its distribution in the head are related to the distance of the radiating element from the skin surface and the current distribution on the antenna. For example, location of the highestSAR typically occurred near the feed point of the antenna. Also, the peak SAR is lower for longer antennas compared to shorter ('),)4) antennas coinciding with the fact that the high current region for the longer antenna is higher up along the antenna and is further away from the surface of the head.

32

Table 6. Computer SAR in anatomic phantoms exposed to wireless communication devices (Normalized to 600 m W output power)

Brain Peak* Frequency SAR SAR Antenna Authors (MHz) (W/kg) (W/kg) Type Date

835 1.10 1.48 0.221 Gandhi et aI.[1994] 0.68 0.74 3/8 1 or Gandhi [1995]

835 0.64 2.17 1141 Tinniswood et aI. [1998] 0.78 1.62 3/8 1

900 1.13 1.37 1141 Gandhi et aI. [1994]

900 <1.38 1.38 Yl1 Dimbylow & Mann [1994]

900 <0.54 0.54 Internal Jensen & Ratmat-Samii <1.20 1.20 1141 [1995] <2.28 2.28 ~1

900 4.69 0.451 Rombach et aI. [1996] 900 1.02 1141 Watanabi et aI. [1996] 915 1.56 2.34 1141 Okoniewski & Stuchly

[1996] 1500 2.28 1141 Watanabi et aI. [1996] 1800 <2.88 2.88 Yl1 Dimbylow & Mann

[1994] 1900 0.77 6.96 1141 Tinniswood et aI. [1998]

0.86 2.74 3/81

*Peak SAR averaged over 1 g of tissue anywhere in the head, including the ear.

At the feed point of the antenna, the current density is the highest which give rise to a strong magnetic field. Because of the dominance of magnetic field coupling in the near zone of the antenna, it is expected that the induced electric field would be closely correlated with the current distribution on the antenna. The maximum SAR and its distribution in the head would follow an eddy current course. Since eddy currents circulating in the ear and the head must go through a narrow appendage, the current density and SAR are the highest at this location. Also, RF electric and magnetic fields are decoupled in the near zone. The effect of electric field would be weaker since dielectric permittivities of muscle and brain tiSSlKs are relatively high at these frequencies. Thus, inductive coupling of antenna current generated magnetic field dominates power deposition in the near zone of a cellular telephone antenna.

Summary

The coupling of RF radiation into biological tissues is influenced by the geometry and composition of the exposed body and the frequency and configuration of the source but also by the

33

Figure 15. Distribution of SAR inside a human head irradiated by a cellular telephone with a vertically oriented A/4 monopole antenna. The peak output power is 0.6 W. [Courtesy ofO. Gandhi, University of Utah].

distance from the source. Both near-zone reactive and far-zone radiative interactions are important for wireless RF personal telecommunication systems. Specifically, in the far zone,

34

Coupling is characterized by plane wave RF field interaction, Electric and magnetic fields are uniquely defined and are related through the intrinsic impedance of the medium, Determination of electric field behavior is sufficient to characterize the interaction, Coupling of RF power from air into planar tissue ranges from about 20 to 60% for the wireless frequencies, The interaction is independent of the source configuration in the far zone, Enhanced coupling can occur at greater depth in bodies with curved surfaces, RF energy is resonantly absorbed by the head at 400 to 1500 MHz. SAR maxima or hot spots may occur near the center of the head,

Coupling of RF energy depends on electric field polarization for elongated bodies whose height-to-width ratio is large, Integrated SAR in the head is similar for a homogeneous or inhomogeneous model, and in the near zone,

RF electric and magnetic fields are decoupled and are not uniform, Beam width is divergent and small compared with the head, Wave impedance varies from point to point in the near zone, RF electric and magnetic fields are in time quadrature, Maxima of electric and magnetic fields occur at different locations, RF radiation in the near zone behaves like a standing-wave field, RF energies are transferred back-and-forth between the radiating antenna and the body, Electric field effect is weaker since dielectric permittivity of tissue is relatively high, Inductive coupling of wireless antenna current generated magnetic field dominates power deposition,

SARs vary with specific antenna configuration and placement of the antenna next to the head,

Anatomy of the head and tissue inhomogeneity can influence the maximum value and distribution of SAR in the head of a mobile telephone user, Integrated SAR in the head is similar for a homogeneous or inhomogeneous model, A large fraction (40-50 %) of the radiated RF power is absorbed by the human body, The bulk of power deposition is on the side of the head nearest to the radiating structure of the cellular telephone, SAR distribution follows an exponential trend away from an antenna side and it is considerably lower elsewhere in the head,

Maximum SAR and its distribution in the head are functions of distance of radiating element from the skin surface and the current distribution on the antenna.

CONCLUSION

The emphasis of this chapter has been a better conceptual understanding of the mechanisms of coupling rather than numerical precision or adherence to anatomic details of the models used. The aim is to gain insights that may be applicable to a wider variety of experimental or exposure situations. While ELF fields and RF radiation all are parts of the same known electromagnetic spectrum, their mode of coupling into biological tissues.and their mechanism of interaction can be quite different. This chapter discussed the coupling ofE'LF field and RF radiation and mechanisms of interaction through the use of a Maxwellian formulation. A major emphasis has been placed on differences between low frequency (especially ELF) and R F energy coupling into biological bodies.

All open electromagnetic systems in which current and charges change in time would radiate electromagnetic energy. An elementary dipole antenna whose length is short compared with a wavelength but large compared with its diameter is employed to illustrate the behavior of such a radiating system. It has been shown that the space surrounding a dipole antenna can be divided into near and far zones as a function of distance from the dipole antenna. The demarcating boundary occurs at a conservative distance ofR = 2D2f) .. , where D is the largest dimension of the antenna. Furthermore, the near zone can be divided into two subregions: the radiative region and the reactive region. In the radiative region, the region closer than 2D21'A., the radiated power varies with distance

from the antenna. In the vicinity of the antenna where the reactive components predominate, it is known as the reactive region. The precise extent of these regions varies for different antennas. For

35

most antennas, the transition point between reactive and radiative regions occurs from 0.2 to 0.4 D2/').. For the elementary dipole antenna, the reactive component predominates to a distance of approximately, ').f21t, where the radiative and reactive components are equal to each other. However, the outer limit is on the order of a few wavelengths or less in most cases. In the far or radiation zone, the electric and magnetic fields have only transverse components. They are outgoing in the radial direction with plane wave fronts.

A typical wavelength in free space at ELF frequencies is 5000 km. The ').f21t distance is about 800 km for the induction and radiation fields to have equal amplitudes. Therefore, for most purposes, ELF transmission line fields are not radiative, but inductive and quasi-static in nature. This fact governs the coupling and induced.field characteristics of ELF and other low frequency electric and magnetic fields in biological tissue. In particular, (l) the electric field is enhanced at the surface of the biological body and is perpendicular to the surface of the body; (2) electric and magnetic fields are decoupled inside a biological body; (3) electric fields applied through air is weakened by about 10-6 upon penetration into biological tissues; (4) magnetically induced electric field encircles the magnetic field and produces an eddy current whose magnitude increases with distance from the center of the body; and (5) eddy currents appear in each region inside the body with a different conductivity which behaves as a unit with its own body center and radius or an equivalent radius. These observations apply to all frequencies where the wavelength is long or the largest dimension of the body is small compared with a wavelength.

In contrast, the 2D2/'). distance is approximately 6 cm for alO-cm RF antenna operating at 900 MHz in free space. Clearly, both near-zone reactive and far-zone radiative interactions are encountered in the vicinity of wireless RF telecommunication systems. In the near zone, the coupling ofRF energy into to the head is substantial. As much as 40-50 % of the radiated RF power is transferred back and forth between the radiating antenna and the body. SAR will vary with specific antenna configuration and its placement next to the head. The bulk of power deposition is on the side of the head nearest to the radiating structure of the cellular telephone and follows an exponential trend away from an antenna side. Anatomy of the head and tissue inhomogeneity can influence the maximum value and distribution of SAR in the head of a mobile telephone user. However, the integrated SAR in the head is similar for a homogeneous or inhomogeneous model. Some of the major features of near zone field are that (1) RF electric and magnetic fields are decoupled and are not uniform, (2) wave impedance varies from point to point, (3) beam width from the antenna is divergent and is small compared with the head, (4) electric field effect is weaker since dielectric permittivity of tissue is relatively high, and (5) inductive coupling of wireless antenna current generated magnetic field dominates power deposition.

In the far zone, coupling is characterized by plane wave RF field interaction and is independent of source configurations. Electric and magnetic fields are uniquely defined through the intrinsic impedance of the medium. Thus, a determination of the electric field behavior is sufficient to characterize the interaction. RF power coupling from air into planar tissue ranges from 20 to 60% for the wireless frequencies. However, enhanced coupling can occur at greater depth in bodies with curved surfaces. In fact, RF energy is resonantly absorbed by the head at 400 to 1500 MHz and SAR peaks or hot spots may occur near the center of the head. The interaction of RF energy with

biological systems depends on electric field polarization for elongated bodies whose height-to-width ratio is large. It is significant to observe that the integrated SAR or total absorption in the head is similar for a homogeneous or inhomogeneous model.

36

REFERENCES

Anderson, V. and Joyner, K.H., 1995, Specific absorption rate levels measured in a phantom head exposed to radio frequency transmissions from analog hand-held mobile phones, Bioelectromagnetics 16: 60-69.

Balzano, Q.,Garay, O. and Steel, F.R, 1978, Energy deposition in simulated human operators of 800 MHz portable transmitters, IEEE Trans. Vehicular Tech. 27:174-181.

Balzano, Q., Garay, O. and Manning, T.J., 1995, Electromagnetic energy exposure of simulated users of portable cellular telephones, IEEE Trans. Vehicular Tech. 44:390-403.

Chen, H.Y. and Wang, H.H., 1994, Current and SAR induced in an human head model by the electromagnetic field irradiated from a cellular telephone, IEEE Trans Microwave Theory Tech. 42:2249-2254.

Cleveland, RF., Jr. and Athey, T. W., 1989, Specific absorption rate (SAR) in models of the human head exposed to hand-held UHF portable radios, Bioelectromagnetics 10: 173-186.

Dimbylow, P.J., 1993, FDTD calculations ofSAR for a dipole closely coupled to the head at 900 MHz and 1.9 GHz, Physics in Medicined and Biology, 38:361-368.

Dimbylow, P.J. and Mann, S.M., 1994, SAR calculations in an anatomically realistic model of the head for mobile communication transceivers at 900 MHz-LS GHz, Phys. Med Bioi. 39:1537-1553.

Durney, C.H., Johnson, C.C, and Massoudi, H. 1975, Long wave-length analysis of plane wave irradiation pf a prolate spheroid model of man. IEEE Trans. Microwave Theory Tech.23:246-

Gabriel, S., Lau, RW, Gabriel, C, 1996, The dielectric properties of biological tissues. 2. Measurement in the frequency range of 10 Hz to 20 GHz. Phys. Med Bioi. 41: 2251-2269.

Gandhi O. P., Gu YG, Chen JY, Bassen ill, 1992, Specific absorption rates and induced current distributions in an anatomically based human model forplane-wave exposures. Health Phys. 63:281-90

Gandhi, 0., 1995, Some numerical methods for dosimetry: ELF to microwave frequencies, Radio Science 30: 161-177.

Gandhi, O.P., Lazzi G. and Furse C.M., 1996, Electromagnetic absorption in the human head and neck for mobile telephones at 835 and 1900 MHz,JEEE Transactions on Microwave Theory & Techniques, 44:1884-1897.

Gandhi O.P., Lazzi G., Tinniswood, A. and Yu Q.S., 1999 Comparison of numerical and experimental methods for determination ofSARand radiation patterns ofhandheld wireless telephones. Bioelectromagnetics. 20/SuppI4:93-101.

Hombach, V., Meier, K., Burkhardt, M., Kuhn, E., and Kuster N., 1996, The dependence of em energy absorption upon human head modeling at 900 MHz, IEEE Transactions on Microwave Theory & Techniques, 44:1865-1873.

Jensen, M.A. and Rahmat-Samii, Y., 1995, EM interaction of handset antennas and a human in personal communications, Proc. IEEE 83:7-17.

Jordan, E.C. and Balmain, K.G., 1968, Electromagnetic Waves and Radiating Systems, PrenticeHall, New York

Kaune, W. T. and Phillips, R. D. 1980, Comparison of the coupling of grounded humans, swine, and rats to vertical, 60 Hz electric fields, Bioelectromagnetics 1: 117-129.

37

Kuster, N. and Balzano, Q.,1992, Energy absorption mechanism by biological bodies in the near field of dipole antennas above 300 MHz. IEEE Trans. Vehicular Tech. 41: 17-23.

Kuster, N. and Balzano, Q., 1997, Experimental and numerical dosimetry. In: Mobile Communications Safety, Kuster N, Balzano, Q. and Lin, J. C, eds., Chapman and Hall, London, UK, pp 13-64.

KusterN, Balzano, Q. and Lin, J. C, 1997. Mobile Communications Safety, Chapman and Hall, London,UK

Lin, J. C., 1986, Computer methods for field intensity predictions, In CRC Handbook o/Biological Effects o/Electromagnetic Fields, C. Polk and E. Postow, eds., CRC Press, Boca Raton, FL, pp.273-313.

Lin, J. C., 1997, Health effects of radiofrequency radiation from wireless communication technology, In: Advances in Electromagnetic Fields in Living Systems, vol. 2, J. C. Lin, ed., Plenum Press, New York, pp 129-164.

Lin, J. C., 1999, Catheter microwave ablation therapy for cardiac arrhythmias, Bioelectromagnetics, 20/Supplement 4: 120-132,

Lin, J.C. and Gandhi, O.P., 1996, Computational methods for predicting field intensity, In: Handbook 0/ Biological Effects 0/ Electromagnetic Fields, C. Polk and E. Postow, eds., CRC Press, Boca Raton, FL, pp 337-402.

Lin, J.C., Guy, A.W. and Johnson, C.C.,1973, Power deposition in a spherical model of man exposed to 1-20 MHz EM fields, IEEE Trans. Microwave Theory and Techniques, MIT 21: 791-797.

Michaelson, S.M. and Lin, J.C., 1987, Biological Effects and Health Implications 0/ Radiofrequency Radiation. Plenum Press, New York.

NCRP, 1981, Radiofrequency Electromagnetic Fields; Properties, Quantities and Units, Biophysical Interaction, and Measurements, Report 67, National Cotmcil on Radiation Protection and Measurements, Bethesda, MD.

Okoniewski, M. and Stuchly MA., 1996, A study of the handset antenna and human body interaction, IEEE Transactions on Microwave Theory & Techniques, 44:1855-1864.

Silver, S., 1949, Electromagnetic Theory, McGraw-Hill, New York. Schmid, T., Egger, O. and Kuster, N., 1996, Automated E-field scanning system for dosimetric

assessments, IEEE Trans. Microwave Theory & Techniques, 44: 105-113. Shimizu, K., Endo, H., and Matsumoto, G. 1988, Visualization of electric fields around a biological

body. IEEE Trans. Biomed Engineering 35: 296·302. Tinniswood, A.D., Furse, C.M., and Gandhi, O.P., 1998, Computation ofSAR distributions for two

anatomically based models of the human head using CAD filed of commercial telephone and the parallel FDTD code. IEEE Trans Ante~ and Propagation 46:829-833.

Watanabe, S., Taki, M., Nojima, T. and Fujiwara, 0., 1996, Characteristics of the sar distributions in a head exposed to electromagnetic fields radiated by a hand-held portable radio, IEEE Transactions on Microwave Theory & Techniques, 44:1874-1883.

38

PRINCIPLES AND HORIWNS OF MAGNETIC RESONANCE IMAGING

Shoogo Ueno 1 and Norio Iriguchi 2

1 Department of Biomedical Engineering, Graduate School of Medicine University of Tokyo, 7-3-1 Hongo, Bunkyo-ku. Tokyo 113 - 0033 Japan

2 Siemens-Asahi Medical Technologies Limited, Tokyo 141 - 8644 Japan

INTRODUCTION

Magnetic resonance imaging (MRI) is a way of making tomographic images of the body non-invasively. Protons in the body can act like tiny bar magnets, with a north pole and a south pole. When an external magnetic field is applied across a part of the body, each little magnet lines up with the external magnetic field. If a radiofrequency (rt) wave is then transmitted into the tissues, some of the magnets are induced by the energy from the rf wave. The rf wave is then turned off, and subsequently the magnets rebroadcast a signal of the same frequency as the original rf wave. A rf coil picks up the signal from the atomic magnets, and a computer can process the signal and reconstruct an image from the signat.

The first successful nuclear magnetic resonance (NMR) experiments were reported by two independent groups in the U.S. [Purcell et al.,1946; Felix Bloch et al.,1946]. In 1952, Block and Purcell received the Nobel Prize for their observations. Since chemical shift phenomena were discovered soon in the 1950s, NMR techniques were developed as important tools for chemical analysis. The early experiments were limited in scope by the relatively poor instrumentation available at that time. In the late 1960s, superconducting magnets were introduced to NMR experiments and revolutionized the scope of NMR together with the emergence of Fouriertransform NMR [Ernst and Anderson, 1966]. Paul Lauterbur [1973] published the first NMR !mage of two tubes of water. By 1980, the clinical evaluation of MRI had begun, and since that time, there have been continuing developments in instrumentation and applications which have led to the present situation. MRI is now established as an important modality in medical practice. The application techniques of MRI have diversified in recent years. Those are, for example, magnetic

Advances in Electromagnetic Fields in Living Systems, Vol. 3 Edited by J.e. Lin, Kluwer AcadernicIPlenum Publishers, 2000 39

resonance angiography (MRA), perfusion and diffusion imaging, and functional nemoimaging (fMRI). Scanning time has been dramatically shortened. Since there are clear attractions in using non-invasive methods for the study ofliving systems, those MRI techniques which investigate the brain and tissue characteristics appear to have a very bright future. Therefore, in this chapter, we overview the MRI techniques, discuss promising applications on the horizon, and introduce novel techniques of tissue impedance MRI of the brain.

MAGNETIC RESONANCE SIGNAL AND RELAXATION TIMES

Electromotive Foree (EMF)

When protons are placed in a static magnetic field Bo, the nuclear spins begin to precess. The phases of spin precessions are primarily governed by uncertainty, and at random. Therefore, the direction of the net macroscopic vector of magnetization Me is expected to be that of the z-axis, the axis of the external static field Bo. When the radiofrequency (rf) field of the magnetic flux density BI' and the Lannor angular frequency ~ which is equal to yBo is applied to the nuclear spins for a duration Lit and in the direction of the x-axis perpendicular to the z-axis, the nuclear spins precess around the x-axis by the angle ()= y BI' Lit, where yis the magnetogyric ratio of the nuclear spins. This occurs because the B.* field is the only apparent field experienced by the

z

Magnetization Me

Flip angle e = rBI * Llt

x

Figure 1. The precession of the magnetization Me around x-axis.

40

nuclear spin in the rotating frame (Fig. 1). Thus the vector of magnetization Mo precesses around the x-axis, and the flip angle "is the phase of the macroscopic precession of the magnetization Mo.

When the rf coil and the sample are supplied with the rf power P, the current ;* of the coil is i" = (P / R) la, where R is the equivalent, gross resistance of the rf coil and the sample. If BI is defined as the magnetic flux density produced by a coil carrying unit current, the magnetic flux density BI* which is actually applied to the sample is BI* = BI (P / R) 112. The BI value of the rf coil can be detennined also by the geometrical configuration of the rf coil.

Figure 2 shows the magnetization Mo· of a volume of interest (VOn of a sample at the time when the flip angle "has been made 90 0 by the rf field (BI cos ~ t) generated by the coil carrying unit current i = cos ~ t. Magnetic resonance signals can be acquired with the same coil, and the electromotive force (EMF) qS induced by the magnetization Mo· is given by

Since the rffield is BI cos ~t, the EMF is exhibited by

The number N* of spins in the VOl is N* = 103 C V A, where A is Avogadro's number (6.02 * 10 23

), V is the volume of VOl in m3, c is the concentration of spins in M Then, Mo' = 103 C V A y2 ( h /

2n") 2 Bo I ( 1+ 1 ) / 3 k T, where h is Planck's constant, I is the spin quantum number, k is Boltzmann's constant, and Tis the temperature in Kelvin. If the concentration c, the rffield BI and the volume v of VOl are detennined, then the qS can be numerically calculated, since al other constants are known.

Main static field Bo

Unit current i

= cos CUo t

Magnetization Mo *

Figure 2. Magnetization Mo· just after the application of a 90 0 pulse.

41

Relaxation Times

Proton density is an important source of the NMR signal that conveys much information about the tissue. But, the total NMR signal has more information than just the number of nuclei reflected in the amplitude of the signal. One can characterize a given tissue by the way the nuclei relax. Relaxation is the term given to a process of transition from an unstable, high-energy state to a stable, low-energy state. When proton spins are exposed to an external magnetic field, the individual magnetic moments align parallel or antiparallel with the external magnetic field. Spins whose expected axes of precessions are parallel with the external magnetic field are called a spins and are in a low-energy, stable state. Spins whose precession axes are antiparallel with the external magnetic field are called fJ spins and are in a high-energy, unstable state. In the state of thermal equilibrium, spins are in only one of those two states, and the number of a spins is slightly greater than the number of fJ spins. As small as this slight excess is, it accounts for the small macroscopic net magnetization Mo directing parallel with the static ma~'11etic field, and it is this differential that accounts for the NMR signal on which MRl is based. When a 90 0 rf pulse is applied, a spins are excited and elevated to the high-energy state, and the number of a spins become the same as that of fJ spins. When a 180 0 rf pulse is applied, the number of fJ spins becomes greater than that of a spins. If excited pins are left in the thermal equilibrium, they will be allowed to lose energy to become low-energy, stable a state. This is the process of Tl or longitudinal relaxation, or spinlattice relaxation. Tl relaxation is an exponential process, and Tl is the time constant of the process. Tl relaxation can be expressed as a macroscopic phenomenon, dMz / dt = (Mo - Mz) / Tt, where Mz is the z-axis component of the transitional magnetization. Tl relaxation is also the macroscopic phenomenon where the magnetization is allowed to return to the initial state of the thermal equilibrium. For example, the proton Tl value of the white matter of the human brain is typically 690 rns at 20 MHz. When molecules are restricted in movement as in solids, Tl relaxation rarely occurs, and Tl value are even as long as several hours. The nuclei are often the elements of large molecules, and the term lattice of spin-lattice relaxation usually means the whole environment of the molecular complex.

Spins have magnetic features affecting each other. Here, the magnetic flux density of the external static field actually experienced by a spin is varied by other spins. For this reason, spins placed in the same static field precess with slightly different frequencies, and phases of spin precessions become incoherent. Thus, T2 or spin-spin relaxation takes place. T2 relaxation is also an exponential process, and T2 is the time constant of the process. T2 relaxation can be expressed as a macroscopic phenomenon, dMxy / dt = - Mxy / T2, where Mxy is the component of transitional magnetization perpendicular to the direction of the main static field. T2 relaxation is also the macroscopic phenomenon where the magnetization decays in amplitude in the ~plane. By T2 relaxation, the amplitude of EMF ~s becomes smaller and exhibits free induction decay (FID). The FID signal ~s can be expressed as

and the initial value of the amplitude ~so is therefore ~so = ~ Bl Mo". In contrast to Tl relaxation, the exchange of energy in T2 relaxation does not take place between spins and lattices but only

42

among spins. For example, the proton T2 value of the white matter of the human brain is typically 1l0ms.

T2 relaxation is accelerated in large molecules, and T2 values of the nuclei of large molecules are generally short. T2 values of nuclei of small molecules are also short when small molecules are combined to large molecules. For example, an anticancer drug fluorourasil solution exhibits a sharp 19F resonance peak and has a longer T2 value than 200 rns, but once it is incorporated into the liver cells, the T2 value becomes shorter than 10 ms.

When the external static magnetic field is inhomogeneous, phases of spin precessions become incoherent. TIris process appears like a relaxation and is called T2' relaxation.

900 pulse !

1800 pulse !

f so (1 - e - T,/T,)

\,.,,-__ ,_, T2 relaxation -"'--- f SE ------------------_._-

.-._._ .. _ ......... -

------ _._._ ... _- t

Echo time

Figure 3. Spin echo (SE) signal.

By shimming, or improving the homogeneity of the main static magnetic field, the T2' value is made long. The FID signal qS from spins in an i:1homogeneous static magnetic field can be expressed as

After spins are excited by a 90 0 pulse, the vector of magnetization Mo on the xy-plane begins to decay in amplitude by T2' relaxation. However, if a 180 0 pulse follows immediately, the phases are refocused to be coherent and the spins generate an echo [Hahn, 1950]. TIris is the spin echo (SE) signal, and the 180 0 pulse is often called a time-reversing pulse, because it refocuses the phases of spin precessions as if time were reversed (Fig. 3). The spin echo (SE) peak signal amplitude qSE is affected by TI relaxation during the repetition time Tr and the T2 relaxation during the echo time Te and expressed as qSE = qSO { 1 - exp (- Tr / T1)} exp (- Te / T2)' The SE signal qS can be hence expressed as

43

The SE signal ;S has a feature where the amplitude is symmetrical in regard to the echo time T Co

Therefore, if the excitation pulse and refocusing pulse are B! and 2~ instead of 90 0 and 180 0

respectively, then the signal ~ can be expressed as

OVERVIEW OF THE SPIN-WARP IMAGING METHOD

In MRI scanners, the scanning space inside the magnet is divided by three intersecting planes corresponding to three anatomical planes of the body (axial, coronal and sagittal). The three planes are indicated by three intersecting axes, with the z-axis indicating the alignment of the main static magnetic field. A gradient magnetic field can be ust;d to assign nuclei to a position in space within the body. The gradient field is generated by a gradient coil [Tumer,1993]. The gradient coils create opposing magnetic fields because they carry currents in opposite directions. The field from one coil opposes the field from the other, creating a net field which serves as a gradient between the two coils. Gradients can be applied in each of the three orthogonal directions (X, y, and z). Slice selection in an MRI subject is the process of recognition of spins precessing in a specific plane perpendicular to an axis of space. The slice selection is performed by applying a gradient magnetic field and an excitation rf pulse simultaneously. Also, the recognition of spin distribution in the 2nd and 3rd directions are perfonned by employing gradient coils. The spinwarp imaging method is the traditional term for popular methods of MRI, such as field echo (FE) imaging, spin echo (SE) imaging and echo-planar imaging (EPI). The main feature of spin-warp imaging method is the process of detecting the location of a spin in the phase of precession by varying a linear, gradient magnetic field [Edelstein et al., 1980].

Reeognition of Spins Distributed in the First Direction

When a sample is placed in a static magnetic field Bo and a linear, gradient magnetic field Gz is superimposed to the static field Bo, spins in th~ sample are excited by a rf field only in the region where the net flux density is specific to the spin resonance frequency ~. This process is the slice-selective excitation of spins, and the selection is in the z-axis direction of the applied gradient field Gz• By modulating the excitation power, the rffield is often given a square envelope fimction for a square excitation, a (lit) sin t envelope function for a sinc excitation, or a Gaussian envelope function for a Gaussian excitation. Since a Fourier transfonned sine function in time is a square function in frequency, the slice profile resulted from a sine excitation is square. The effective frequency bandwidth of an excitation is the reciprocal of the effective duration of the applied rf pulse. When an excitation pulse has a sinc envelope function with a main lobe of I ms in duration at half maximum, the resulting slice has a square profile of about I kHz in bandwidth. The smaller the excitation bandwidth is, the thinner the resulting slice thickness is. If the frequency bandwidth is fixed and unchanged, the slice thickness in the z-axis direction becomes thinner with a larger gradient field Gz•

44

RF field fo = 42,600,000 + 4,260 [Hz]

Pulse width at half maximum

2.35 [ms]

10mm mT/m

'-----+ z

--+l j.- Slice thickness 1 mm

Figure 4. Slice selection with a gradient field.

45

As shown in Fig. 4, when the gradient field Gz is 10 mT/m, the proton resonance frequency is made distributed in 426 Hzlmm, because the resonance frequency of protons is approximately 42,600,000 HzIT. When a pulse is applied with a sinc envelope with a main lobe of 2.35 ms in duration at half maximum, the Fourier transfonned bandwidth is 426 Hz. Hence, a slice is selectively excited with a thickness of 1 mm. If the frequency fo of the rf pulse is fo = 42,600,000 Hz + 4,260 Hz, then the position of the slice selection will be shifted by 10 mm in the + z-axis direction.

Recognition of Spins Distributed in the Second Direction

The recognition of spins distributed in the 2nd, or the x-axis direction can be perfonned by acquiring signals simultaneously applying a linear, gradient magnetic field Gx in the x-axis direction. With the gradient field G-x. the resonance frequency of spins distributed in the x-axis direction is linearly distributed in the direction of x-axis from lower frequencies to higher frequencies than that of spins without the gradient field. For example, if the gradient field Gx is 10 mT/m, then in the region of - 0.2 m to + 0.2 m inclusively in the x-axis direction, the frequency deviation of resonance will be linearly distributed from - 8,400 Hz to + 8,400 Hz inclusively. This frequency bandwidth is small enough to be processed by an ordinary computer. The signal acquired by a rf coil has various frequencies specific to the locations of spins distributed in the xaxis direction, and the spatial distributions of spins are restored by Fourier transformation. The process of encoding spin locations is simply called reading-out of the signal. By the Fourier transfonnation, a projection curve of the sample in the x-axis direction can be obtained. The projection data are often exhibited in an absolute value (142 + 1m2) III with the real amplitude 14 and imaginary amplitude 1m, both detected by quadrature demoduration. For example, if the number of acquired data points is 512, and the half of those points are real and the rest are imaginary, then the projection can be exhibited in a 256-point curve. With the application of a reading-out gradient field Gx. the signal intensity rapidly diminishes. This is because spin frequencies are made distributed in the x-axis direction and the phases of spin precessions rapidly become incoherent This decay can be altered by applying a negative gradient field - Gx and then reversing the gradient field to a positive + Gx• By this, a field echo (FE) signal is generated. As

shown in Fig. 5, the FE is generated at the time when the gray areas are the same. When the gradient field Gx is 1 mT/m, proton frequency deviation occurs due to the fact that the gradient field Gx is 42,600 Hzlm in the x-axis direction. If the field of view (FOV) of the MR image is, for example, 32 em in the x-axis direction, then the frequency bandwidth covering the FOV range of

,FE

_ ... _.~E __ ._~ .. _ .. _ ..... _

~ ts ---+I

Figure 5. Reading-out gradient field Gr and field echo (FE).

46

--------------------~~--------------------

t [8]

Figure 6. Echo signal acquired with the reading-out gradient field Gr.

f [liz]

Figure 7. Fourier transformed real part of the echo.

f [liz]

Figure 8. Fourier transformed absolute value of the echo.

47

32 em will be 13,600 Hz, and according to the Sampling Theorem, the sampling rate of the AD converter must be greater than 27,200 Hz. In spin echo (SE) imaging, most of acquired signals are the combined echoes of the SE signal generated by a 180 0 pulse and the field echo (FE) signal generated by reversing the reading-out gradient field Gr. UsualIy, the moment of SE refocusing and the moment of FE refocusing are made coincident. The acquired data contains various frequencies of spins, and the frequencies are restored to the original state of spin locations by the Fourier transfonnation.

Figure 6 is an example of an echo signal acquired during the time fs• The axis is time in sec. Figures 7 and 8 show the real part Re and absolute valueAv = (Re2 + 1m2) 112 of the signal data. The axis is frequency in Hz. The curve of absolute value Av is similar to the envelope shape of the real part Re cmve.

Recognition of Spins Distributed in the Third Direction

The recognition of spins distributed in the 3rd direction is perfonned by a technique called phase-encoding. It begins after the slice-selective excitation. A smalI gradient field Gy in the 3rd, y-axis direction is applied for a certain period LIt. By the application of the gradient field Gy, a spin located far in the + y direction from the original point is exposed to a higher field than that of the original point and precesses at a higher frequency than that of the spin located at the original point. In the point LIt, the phase of spin precession becomes larger than that of the spin at the original point On the other band, a spin located far in the - y direction from the original point, is exposed to a lower field than that of the original point and precesses at a lower frequency. In the period LIt, the phase of spin precession becomes smalIer. Secondly, slightly larger gradient filed is applied in the same y direction for the same period LIt and also after the second slice-selective excitation. Thus, the gradient field is successively changed stepwise in amplitude and applied, for example, in 256 steps. By this, the farther spins are located from the original point, the greater

changes of gradient field in amplitude are experienced by the spins. By Fourier transfonnation of the acquired 256-data set, it is recognized how far the spins are located from the original point

When the gradient field Gy is varied in amplitude 256 times, for example, from - 1.27 mT/m to + 1.28 mT/m by the step of 0.01 mT/m, a spin at the point 10 em apart from the original point in the y-axis direction, experiences a deviation of the field in amplitude by 0.001 mT each

RF~

Figure 9. Field echo (FE) imaging sequence.

48

y-axIS phase (3 [Tad / step]

frequency f [Hz] x-axis

Figure 10. Frequency-phase plane.

---LJ L

Figure 11. Spin echo (SE) imaging sequence.

49

time, and the proton resonance frequency changes 256 times by 42.6 Hz each time. With the duration LIt of 3 ms, at the point 10 cm apart from the original point, the phase of the spin changes 256 times by 2 1r * 42.6 * 0.003 = 0.8 rad every time. By Fourier transformation, the spin distribution in the y-axis direction is restored. A typical field echo (FE) imaging sequence is, as shown in Fig. 9, a combination of a slice-selective excitation in the z-axis direction, a frequencyencoding in the x-axis direction and a phase-encoding in the y-axis direction. By two-dimensional Fourier transformation of a set of acquired signal data, the two-dimensional distribution of spins is restored. Hence, the resulting image is on the plane of one axis in frequency f in Hz and another axis in phase (J in radlstep (Fig. 1 0).

A spin echo (SE) imaging sequence employs a 180 0 pulse (Fig. 11). The 180 0 pulse is applied in the period Te / 2 after the 90 0 selective excitation. The dephased spin phases refocus to generate an echo signal in the echo time Te.

k- Space

By varying the amplitude of the gradient field in the phase-encoding direction, spins are labeled according to their phases of precessions. The successively acquired data are then put in successive order on a plane traditionally called k-space [Ljunggren, 1983]. The echo signals put on the k-space are those demodulated and subtracted by the primary frequency ~ = 21r I-tl. Obviously, each signal data appears to have waves containing different frequencies. Those waves are specific to the horizontal x-axis distribution of spins and generated by the reading-out gradient field Gr.

The waves in the vertical direction are also specific to the vertical y-axis distribution of spins and generated by varying the amplitude of the phase-encoding gradient field Gp• When a quadrature demoduration is employed, each data has a real part and imaginary part. For example, if the number of data points of a signal is 512, then 256 m-e real data points and the remaining 256 are imaginary data points. Hence, when the number of phase-encoding steps is 256, the vertical,

Figure 12. Acquired real part data on the k-space.

50

Figure 13. One-dimensional Fourier transformed real part data on the k-space.

51

Figure 14. One-dimensional Fourier transfonned absolute value data on the k-space.

52

Figure 15. Two-dimensional Fourier transformed absolute value data on the k-space.

phase-encoded data have 256 real data points and 256 imaginary data points. Therefore, on the kspace, there are a real data set of256*256 points and an imaginary data set of256*256 points. Fig. 12 shows an example of a real data set of 256*256.

Figure 13 shows 256 one-dimensional Fourier transformed real Re data on the k-space. Each horizontal data consists of 256 broken lines. The vertical array is not Fourier transformed. Fig. 14 shows one-dimensional Fourier transformed absolute value Av = ( R/ + 1m2 ) 112 data on the k-space. Each horizontal data consists of 256 broken lines, while the vertical array is not Fourier transformed.

A two-dimensional spin distribution of an orange is shown in Fig. 15. This was restored by a two-dimensional Fourier transformation of the horizontal and vertical waves on the k-space. The spin distribution is displayed in absolute values. The matrix size is 256*256. The unit of the horizontal axis is frequency fin Hz and corresponds linearly to the distance I in m. The unit of the vertical axis is phase () in rad/step and also corresponds linearly to the distance I in m. If the intensity diagram is gray-scaled and displayed on a CRT, then we obtain an ordinary MR image.

Image Contrast

By applying an excitation rf pulse, the magnetization vector is given a component perpendicular to the z-axis and the component decays in amplitude by T2 * relaxation. When a field echo (FE) image sequence is used, and if both the repetition time Tr and echo time Te are long, then the resulting FE image is a T2*-weighted image affected by T2* relaxation. The T2* relaxation of the brain tissue often reflects the brain activity, and T2 * -weighted images are important in the field of functional neuroimaging (fMRl).

With a spin echo (SE) imaging sequence, the magnetization vector is given a component perpendicular to the z-axis and the component decays in amplitude by T2 relaxation. If the repetition time Tr is longer than Tl and the echo time Te is shorter than h then the resulting SE image reflects the spin density distribution.

53

If the echo time Te remains short and the repetition time Tr is shortened, the sample of long T, undergoes successive excitations before the magnetization recovers by T, relaxation and generates small signal. The resulting image is a T,-weighrted image and has a constant reflecting the T, distribution. If both the repetition time Tr and echo time Te are long, then the resulting SE image is a hweighted image affected by T2 relaxation. Compared with field echo (FE) images, the influence of inhomogeneity of the main static field is reduced in SE images.

Figure 16 is an example of inversion recovery, turbo spin echo (IR-SE) image of the brain obtained at 1.5 T. The time 1; denotes the duration from an inversioin 180 0 prepulse application through the rapid SE signal acquisition. With the long 1;, the image is primarily T,weighted. However, the image is also slightly Trcontrasted because of the long Te. Fig. 17 is an example of turbo gradient spin echo (TGSE) image of the brain in 1024*1024 matrix, obtained in 246 sec at 1.5 T. With the long Te, the image is primarily hweighted. Also, field echo (FE) signals have been inserted into the k-space, and the image quality is improved in anatomical details.

Figure 16. Inversion recovery, turbo spin echo (IR-TSE) image of the brain obtained at 1.5 T (Tr = 4,500 rns, 1; = 450 rns, Te = 60 ms). (Siemens AG / Siemens-Asahi Medical Technologies Ltd.)

T,-weighted images give higher intensity signals from the white matter than the gray matter. T2-weighted images give higher intensity signals from the gray matter than the white matter. The contrast arises because the water protons of white matter have shorter T, and T2 values than those of gray matter. The shorter T, value causes the white matter protons to undergo less saturation, and hence, to generate higher signal intensity in the T,-weighted image. On the other hand, because of their shorter T2 value, the white matter protons undergo a greater degree of signal loss during the echo time Te of the T2-weighted image.

54

Figure 17. Turbo gradient spin echo (TGSE) image of the brain in 1024*1024 matrix obtained at 1.5 T (Tr = 6,000 ms, Te = 115 ms). (Tokyo Metr. Ebara Hospital / Siemens-Asahi Medical Technologies Ltd.)

55

DIVERSIFICATION OF MRI APPLICATION TECHNIQUES

Magnetic Resonance Angiograpby (MRA)

Magnetic resonance angiography (MRA) is developing a powerful technique to obtain flow images. It does not necessarily require the injection of a contrast agent and pennits the measurement of flow velocities. It is becoming increasingly used for the assessment of abnormalities in the cerebrovascular system and peripheral vasculature.

There are two important methods for MRA. Those are the time-of-flight (TOF) method and the phased-contrast (PC) method. In a TOF method. the water protons in a slice of a certain thickness are first labeled. The labeling is performed by applying an excitation rf pulse. Since saturation makes the numbers of a and f3 spins equal, few signals are generated from the saturated region. In a time t after the saturation, fresh and unsatwated spins flow into the saturated region. As a result, contrast is generated between stationary and flowing water protons. Threedimensional, time-of-flight (TOF) MRA is currently used in clinical practice. If the consecutive excitation pulses are applied at intervals that are much shorter than h there is an increase in signal intensity, because the inflowing spins generate more signals than the stationary spins, which undergo partial saturation. Gradient-echo, FLASH (fast low-angle shot) sequences are commonly used for the 3D, time-of-flight MRA. The 3D data sets contain information about both stationary and flowing spins. Vascular structure can be extracted out of the full data set by virtue of their increased intensity via a postprocessing routine called maximum intensity projection (MIP) (Fig. 18).

Flow can induce phase changes in MR signals. As water protons move along a gradient magnetic field, they accumulated a phase shift that is proportional to their velocity. In the phasedcontrast (PC) method. a gradient field is applied in a certain direction. The faster the spins flow in the gradient direction, the more the transverse magnetization changes phase. By subtraction of data sets with and without the gradient field, an angiogram is obtained. Three-dimensional, PC

Figure 18. Turbo-MRA of the brain. (Tokyo Metr. Ebara Hospital / Siemens-Asahi Medical Technologies Ltd.)

56

MRA is currently used clinically. Compared to the TOF method, the PC method takes longer for signal acquisition, but the background tissue signals are small.

Perfusion and DitJusion Imaging

A method called arterial spin labeling is used in visualizing perfusion. Once spins in the neck region are excited, arterial flow will carry the labeled spins into the brain tissue and influence the observed signal from brain water. Subtractions of images with and without the spin labeling permit measurements of regional perfusion [Edelstein et al. 1980; Ljunggren, 1983]. A technique of signal targeting with alternating radiofrequency (EPIST AR) is also used [Williams et al., 1992; Zhang et al., 1993]. This is an echo-planar imaging (EPI) technique combined with time-of flight (TOF) techniques, and generates a perfusion image by subtraction of images with and without saturation prepulses [Edelman et al., 1994]. The EPI is a rapid imaging technique to obtain a 2-dimensional data set by filling the k-space with field echo (FE) signals generated by an excitation pulse.

Diffusion measurements by NMR were introduced in the 1960s [Stejskal, 1965; Stejskal and Tanner, 1965]. In water, all molecules move randomly. It is often referred to as Brownian motion, and involves displacement distances that are small and comparable to cell sizes. Hence, the water diffusion can be affected by the cell structure and diffusion-weighted images can reflect the change of the cell structure. In the tissue of cerebral infarction, diffusion of water is restricted and the apparent diffusion coefficient (ADC) of water becomes small. This is allegedly caused by the water shift from the extracellular spaces into the intracellular spaces [Mosley et al. 1990; Kohno et al., 1995]. Spins moving randomly under a gradient field lose signals, and the diffusion can be detected by MRI. Neither a TI- nor T2-weighted image reveals the lesions of cerebral infarction obtained in a few hours after the onset, but a diffusion-weighted image often reveals the lesions in the early stage of infarction [Mintrovich et al., 1991; Pennock et al. 1994; warck et al" 1992 ]. Diffusion-weighted images are often linked to the anisotropy of the tissue structure. The anisotropy can be detected by the relations between the direction of the gradient field and the extent of signal loss by the water movements [Le Bihan and Turner, 1992]. Associated with a cerebral infarction, the swelling of cells and shrinkage of extracellular spaces can lead to the reduction of electric conductivity in the early stage of infarction [Verheul et al., 19941. This suggests the importance of a non-invasive means to measure the tissue impedance of the brain. Fig. 19 shows a T2-weighted, turbo spin echo (TSE) image (upper) and a single-shot, diffusionweighted, echo-planar image (EPl)(lower). Both are images of the same patient 2 hours after the acute stroke. The lesion is not recognized in the T2-weighted image (upper), while the lesion is disclosed in the diffusion-weighted image (lower).

Functional Neuroimaging (fMRI)

Functional neuroimaging (fMRI) is one of the most important MRI techniques developed in recent years and visualizes the active regions of the brain non-invasively. Functional imaging uses field echo (FE) imaging sequences of long echo time Te as 50 ms and reflects T2· characteristics of the tissue. Electron spins of Fe2+ of deoxy-hemoglobin (Deoxy-Hb) have much greater magnetic momentum than those of proton spins, and proton spectra of protein of DeoxyHb solution exhibit broad peaks caused by the effects of paramagnetic features of Deoxy-Hb. Also, the water resonance frequency is broadened by the effects of the gradient fields due to the

57

Figure 19. The acute stroke lesion is not recognized in the T2-weighted image (upper). The lesion is disclosed in the diffusion-weighted EPI (lower). (University o/Tokyo Hospital / Siemens-Asahi Medical Technologies Ltd.)

58

Fe2+ electron spins. Therefore, T2*-weighted images of water arOlmd Deoxy-Hb generate few signals. In the living brain, the gradient fields caused by the Fe2+ electron spins strongly influence not only the water in the blood vessels but also, to much greater extent, the outer neural tissues of the water volume of the blood vessels [Ogawa and Lee, 1990; Ogawa et al., 1990; Turner et al., 1991].

Associated with brain activity, the blood flow increases in the region, oxy-hemoglobin (Oxy-Hb) increases, and Deoxy-Hb decreases in relative quantity. From the blood oxygenation level development (BOLD) effects, the signal intensities from the active region in the obtained T2*-weighted images increases [Frahm et al., 1992; Bandettini et al., 1992]. Also, functional MRI reflects internal verbal generations (Fig. 20) and recalling of sensory or motor stimulation [McCauthy et al., 1993; Ruechert et al., 1994; Le Bihan et al., 1993].

Figure 20. Three-dimensional functional image of the brain obtained during the internal speech word generation.(University of Minnesota / Varian Associates Inc. / Siemens-Asahi Medical Technologies Ltd.)

Magnetic Resonance Spectroscopy (MRS)

The ability to obtain localized spectra offers great potential for detecting human disease at the early stage of biochemical alteration. The proton signal from water is much greater than signals from metabolites. However, the suppression of water is expanding the use of MRI in metabolite evaluation. In the field of magnetic resonance spectroscopy (MRS), metabolite maps of specific chemicals or chemical shift images (CSl) are also obtained. Fig. 21 shows the spectral map (lower right) of the volume of interest (VOl) positioned in the brain (upper left and lower left). The spectrwn (upper right) is from one voxel and featured with the three peaks of choline containing compounds, creatinine, and N-acetylaspartate (NAA).

IMAGING OF IMPEDANCE DISTRIBUTION OF THE BRAIN

Since electrical properties are important charackristics of living organisms, techniques for impedance tomography used to visualize impedance distribution have been developed with

59

Figure 21. Proton spectrum (upper right) from one voxel of the brain and spectral map (lower right) of the volume of interest (VOl) positioned in the brain (upper left and lower left). (UCSF /

Siemens-Asahi Medical Technologies Ltd.)

60

great interest [Metherall et al., 1996]. The previously proposed techniques require that electrodes are attached to the surface of the human body. Meanwhile, MRI techniques have become important tools in medicine and biology. Conventional MRI, however, produces no information about the electrical properties of the body. Therefore, this section addresses the feasibility for visualization of electrical impedance distributions of the brain, based on MRI techniques. In MRI, when conductive tissues are subjected to an excitation rf field, eddy currents are induced in the tissues. The eddy currents generate a reactive field to reduce the applied rf field. The basic idea is to use shielding effects of induced eddy currents in the rody on spin precession. Two types of methods are described [Ueno and Iriguchi, 1998]. One method for visualization of the conductivity distribution of living organisms is to use very large flip angles. The method is used to

obtain conductivity-enhanced MR images at the given Larmor frequency. Another method is to apply an additional time-varying magnetic field parallel to the main static field Bo. The magnetic field is produced by the third coil, named "Be" coil. The method is used to enhance the conductivity in MR images at an arbitrary frequency. Experiments have been carried out to verify these concepts using a 7.05-T, 18.3-cm system [Ueno and Iriguchi, 1998].

Principles

When conductive tissues are exposed to rf magnetic field, eddy currents are induced in the conductive tissues, resulting in the reduction of the net rf fields into the tissues (Fig. 22).

i Static field Bo

Sample Reactive field

_ RF field Bl _ ...........

I

RF coil

The main characters of eddy currents in the tissues are in principle movements of charged particles and rotations of electrical dipoles [Gadian and Robinson, 1979; Bomsdorf et al., 1988; Ong et al., 1995]. The intracellular concentratio<1 of potassium ions within tissues is typically about 150 mM, while the extracellular fluid contains about 150 roM sodium chloride. The conductivity of the tissue samples causes rf losses. Meanwhile, dielectric water dipoles rotate at the Lannor frequency. Dielectric resonance also occurs in pure water and produces macroscopic

61

currents at the Lannor frequency. Also, the dielectric resonance causes rf losses. The shielding effect caused by the eddy currents affects the spin dynamics of nuclei in the tissues. The flip angles are reduced in varied degrees, depending on the electrical characteristics of the tissues, by the shielding effects. Since MR signals are proportional to the sine wave function of the flip angles, the tissue signals can be varied by the shielding effects, reflecting the conductivity distribution of the tissues. Due to the absence of the transversal components of magnetization, when a precise 180 0, 360 0 or 540 0 excitation pulse is applied to conductive tissues, the tissues do not yield a signal. Meanwhile, the resistive tissues yield signals as they are less shielded than conducting tissues and undergo different flip angles simultaneously. Also, the resistive tissues leave transversal components with magnitudes determined by the sine wave functions of flip angles. The difference in signal, therefore, reflects the conductivity of tissues. By applying very large flip angles, conductivity-enhanced MR images can be obtained, at only the given Lannor frequency, and in the direction perpendicular to the applied rf field.

To obtain conductivity-enhanced images at an arbitrary frequency, an additional timevarying field parallel to the main static field Bo is introduced. The purpose of the time-varying field is to perturb the main static Bo field. The perturbing field is produced by the third coil hereby called "Be" coil (Fig. 23). By the perturbing field or "Be" field, the slice positioning of the image is affected, and the slice selection fluctuates according to the function of the "Be" field. Also, the spatial information in the reading-out and phase-encoding directions is affected. Conducting tissues are less affected by "Be" field, because of the shielding effects. Since the frequency of "Be" field is independent on the given Lannor frequency, the conductivity-enhanced images are obtained at any frequency but in the direction perpendicular to the "Be" field.

Materials and Methods

EXperlIDents were first performed on a di~ned water phantom with a specific conductivity (j less than 0.1 Sim and a saline solution phantom with (j of approximately 1.2 S/m.

t Static field Bo r---_

RF fieldB l

" Be " field

Figure 23. The "Be" coil produces the "Be" field parallel to the main static field Bo.

62

Both phantoms were columnar, 2 em in diameter and 4 em long. A conventional spin echo (SE) sequence was used to obtain MR images. By varying the excitation rf power level stepwise, image projections in the reading-out direction were obtained. A 7.05-T, 18.3-cm machine was used. The proton Lannor frequency of magnetic resonance at 7.05 Tis 300 MHz. Excitation pulses were 2-ms Gaussian. To eliminate the longitudinal T\ relaxation effects, the repetition time Tr was set as Tr = 10 sec. Next, a series of image projections of the head of a 4-week old mouse were obtained. A 180 ° SE image of the head was obtained by applying excitation pulses, which were selective to the cerebrospinal fluid (CSF). A four-turn, solenoidal coil of 6 em in diameter was fabricated as the "Be" coil to produce a low-frequency sinusoidal magnetic field of 10 mT in amplitude. The amplitude of the field was comparable to the amplitude of the reading-out gradient field and the maximum phase-encoding field. An 8-week old rat was subjected to imaging with a constant "Be" field of sinusoidal 1 kHz.

Saline .OWater Signal intensity \ '

lA1Al!1111111~~1~ '--_____ + Excitation power

Figure 24. Series of image projections of water and saline solution phantom obtained with excitation power increased stepwise from left to right.

Results and Discussion

Figure 24 shows a series of image projections obtained with the excitation power increased stepwise from left to right. The larger projections, to the right of each image, are those of the water phantom. To the left are smaller projections of the saline solution phantom [Ueno and Iriguchi, 1998]. The saline solution phantom contained air as a marker. The maximum projection intensity resulted from a 90°, 270° or 450° excitation, and the minimum projection intensity resulted from a 180° or 360° excitation. It should be noted that the saline solution phantom requires more power for excitation than does the water phantom. Fig. 25 is a sagittal SE reference image of the mouse head obtained with 90 ° excitation pulses of repetition time Tr of 3,000 ms and echo time Te of 10 ms [Ueno and Iriguchi, 1998]. The horizontal (Fl) axis denotes the phase-encoding direction and the vertical (F2) axis denotes the reading-out direction.

63

Figure 25. Proton density reference image of the mouse head.

Figure 26 shows a series of image projections of the same mouse head obtained in the reading-out (F2) direction with the excitation power increased stepwise from left to right [Ueno and Iriguchi, 1998]. Fig. 27 is the head image obtained with an excitation flip angle of 180 0

selective to the brain cerebrospinal fluid (CSF) [Ueno and Iriguchi, 1998). All scan parameters were identical to those of the reference image (Fig. 25) but different in the excitation flip angle.

Figure 27 shows a slight signal from the brain and muscle tissues because there were few transversal components of magnetization. On the other hand, in the same image, the resistive tissues (e.g., fatty tissues, spinal discs, and ligaments), transparent to the rf field, yielded specific signals. By applying 180 0 pulses to the conducting CSF and muscle tissues, resistive fatty tissues were simultaneously given different flip angles from 180 0 and produced image signals. Fig. 28 is a sagittal, FLASH (fast low-angle shot) reference image of the rat head, and Fig. 29 is the reference image subtracted by a continuous "Bc"-applied image [Ueno and Iriguchi, 1998).

Figure 26. A series of image projections of the mouse head obtained with the excitation power increased stepwise from left to right.

64

Figure 27. Image of the mouse head obtained with an excitation flip angle of 180 0 selective to

CSF.

Figure 28. FLASH reference image of the rat head.

65

The signals of Fig. 29 are mainly from the spinal discs, ligaments, and fatty tissues. Using a continuous "Be" field, more spins were perturbed in resistive tissues than in conducting tissues. The image signal from resistive tissues fluctuated and then disappeared. The potential blurring effects of the "Be" field in the reading-out direction and phase-encoding direction can be eliminated by triggering the "Be" field. Eddy currents in the tissues consist of conduction currents and dielectric currents.

Figure 29. Reference image subtracted by a "Be"-applied image.

It can be readily shown that the shielding effects caused by conduction currents are most significant in the geometric center of a sample. However, dielectric currents exhibit specific characters in different ways. When dielectric resonance is not present, the signal intensity I(r) of a

spin echo (SE) image (~ - 2 ~) obtained with a transmit-receiving coil is expressed as: I(r) = p

fX..r) sin~(r) sin2~(r) {l - exp ( - 1'r / TI)} exp ( - Te / T2), where r denotes the vector of position

and the flip angle fX..r) works as the transmit-receiving function. When dielectric resonance is present, the signal intensity I(r) is expressed as:

I(r) = p fX..r) sin{ ~(r) d(r) }sin2 {~(r)d(r)} {l - exp ( - Tr / TI)} exp ( - Te / T2)

where d(r) is a function of the extent of the dielectric resonance in the effective flip angle

reduction. The effective flop angle ~(r)d(r) is yielded by the ratio of the signal intensities (h II)

of two SE images obtained with two flip angles (2~, ~) respectively: ~(r)d(r) = arccos {her) / 2 II(r)}. Here, cos{~(r)d(r)} is given by her) / 2/1(r), and sin2{~(r)d(r)} = 1 - cos2{~(r)d(r)}. If

two SE image data II(r) and Io(r) are obtained with 1{ / 4 and a very small flip angle excitations respectively, the extent d(r) of dielectric resonance in flip angle reduction can be expressed as

66

follows:

With long-Tr and short-Te scan parameters, a map of the d(r) of dielectric resonance can be produced. Fig. 30 shows a contour distribution map and the central trace curve of the extent of the flip angle reduction. The flip angle reduction was caused by dielectric resonance in the equator

plane of a spherical phantom of distilled water of a relative permittivity hi- of approximately 80 and 35 mm in diameter. It should be noted that the flip angle reduction has a specific coaxial and parabolic distribution pattern and is more significant at the surface region of the phantom and less significant in the geometric center. Since the phantom contains few ionized particles and only electrical water dipole, the distribution pattern is solely due to the rotations of the water electrical dipoles. Since the flip angle reduction reflects the rf power dissipation, the distribution map of the dielectric currents in the distilled water can be derived from the flip angle reduction map. The dielectric currents also exhibit a coaxial yet linear distribution pattern. Living tissues have ionized particles to produce conduction currents and water electrical dipoles those produce dielectric currents. Both give rise to the shielding effects.

Investigation of imaging of the impedance distributions of the brain based on noninvasive MRI techniques is important in brain pathology. It is well known that diffusion-weighted MR imaging is remarkably sensitive to early events associated with cerebral ischaemia.16

Diffusion results from the thermal-translational motion of molecules. It involves displacement distances comparable to cellular dimensions, raising the possibility that the measurement of water diffusion might provide a means of probing cellular integrity and pathology. The diffusionweighted imaging changes are more concretely related to changes in the cell volume, with a shift of water from a relatively unrestricted diffusion environment in the extracellular space to a relatively restricted intracellular environment [Mosley et al., 1990; Kohno et al., 1995]. An investigation with electrodes on a model of the rat brain injury shows that the time course of the intensity changes in the diffusion-weighted images parallel the progressive shrinkage of the extracellular space as measured by the electrical impedance [Verheul et al., 1994]. There are prior studies on Bl field mapping [Hornak et al., 1988; Insko and Bolinger, 1993; Stollberger and Wach, 1996], but those studies have not specifically targeted impedance distribution based on MRI techniques without electrodes. Both the large flip angle method and the "Be" field method are sensitive to rf inhomogeneity. In addition. both the Larmor frequency Bl field and the low frequency "Be" field are easily transmitted, absorbed, and reflected by biological tissue boundaries in varying degrees. The variance depends on the geometry of the subject and the direction of the Bl or "Be" field. However, in spite of these difficulties, the methods of this study are promising to obtain tomographic images of electrical impedance distributions of living organisms.

CONCLUDING REMARKS

Before the mid 1990s, MRI scanners were classified simply by their field strengths, namely, low-field (0.5 T and below), mid-field (1.0 n, and high-field (1.5 T and higher) scanners.

67

Flip angle reduction (arbitrary unit)

F2 (cm)

1

2

3

4

5

6

7

B 7 6 543

Fl (cm)

2 1 o

Figure 30. Contour distribution map and the trace curve of the extent of the flip angle reduction in the equator plane of a spherical phantom of distilled water.

68

Continued progress in high-perfonnance gradient coil technologies have made high-quality and fast imaging a reality. In the last decade, techniques have been diversified dramatically. In the next decade, with the increase of utilization and deeper integration into clinical process, MRI scanners will be classified in a more sophisticated way. That will be by productivity in scanning reality and real-time postprocessing procedures and friendliness in uesign concepts for clinical use. In the field of basic technology, it is apparent that MRI is not simply a method of obtaining radiological images, it can also provide a great deal of physiological information. Some MRI techniques will be developed for non-invasive mapping of physical, chemical and physiological properties such as temperature, mechanical elasticity, pressure, electrical conductivity, permittivity, and metabolite concentration. Therefore, in this chapter we have reviewed the MRI basic techniques and introduced novel techniques of tissue impedance MRI of the brain.

REFERENCES

Bandettini, P.A., E.c. Wong, RS. Rinks, R.S. Tikofsky, andJ.S. Hyde, 1992, Time course EPI of human brain function during task activation. Magn. Reson. Med. 25: 390.

Bloch, F., W.W. Hansen, and M. Packard, 1946, The nuclear induction experiment Phys. Rev. 70: 474.

Bomsdorf, H., T. Helzel, D. Kunz, P. Roeschmann, O. Tschendel, andJ. Wieland, 1988, Spectroscopy and imaging with a 4 Tesla whole-body MR system. NMR Biomed. 1: 151.

Edelman, RR, B. Siewert, D.G. Darby, V. Thangaraj, A.c. Nobre, M.M. Mesulam, and S. Warach, 1994, Qualitative mapping of cerebral blood flow and functional localization with echo-planar MR imaging and signal targeting with alternating radiofrequency. Radiology 192: 513.

Edelstein, W.A., J.M.S. Hutchison, G. Johnson, and T. Redpath, 1980, Spin warp NMR imaging and applications to whole body imaging. Phys. Med BioI. 25: 751.

Ernst, R.R and W.A. Anderson, 1966, Application of Fourier transform spectroscopy to magnetic resonance. Rev. Sci. Instrum. 37: 93.

Frahm, J., H. Bruhn, K.D. Merboldt, and W. Haenicke, 1992, Dynamic MR imaging of human brain oxygenation during rest and photic stimulation. J. Magn. Reson. Imag. 2: 501.

Gadian, D.G. and F.N.H. Robinson, 1979, Radiofrequency losses in NMR experiments on electrically conducting samples. J. Magn. Reson. 34: 449.

Hahn, E.L., 1950, Spin echoes. Phys. Rev. 80: 580. Hornak, J.P., J. Szwnowski, and RG. Bryant, 1988, Magnetic field mapping. Magn. Reson. Med.

6: 158. Insko, E.K. and L. Bolinger, 1993, Mapping of the radiofrequency field. J. Magn. Reson. A 103:

82. Kohno, K., M. Hoehn-Berlage, G. Mies, T. Back, and K.A. Hossmann, 1995, Relationship

between diffusion-weighted MR images, cerebral blood flow, and energy state in experimental brain infarction. Magn. Reson. Imag. 13: 73.

Lauterbur, P.C., 1973, Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature 242: 190.

69

LeBihan, D. and R Turner, 1992, Diffusion and perfusion. in: Magnetic Resonance Imaging Mosby Year Books, St Louis

LeBihan, D., R Turner, TA Zeffiro, CA Cuenod, P. Jezzard, and V. Bonnerot, 1993, Activation of human primaJy visual cortex during visual recall: an MRI study. Proc. Natl. Acad Sci. USA. 90: 1802.

Ljunggren, S., 1983, A simple graphical representation ofFourier-based imaging method. J.

Magn. Reson. 54: 338. McCauthy, G., AM. B1amire, D.L. Rothman, R Gruetter, and RG. Schulman. 1993, Echo-planar

MRI studies of frontal cortex activation during word generation in humans. Proc. Natl. Acad. Sci. USA. 90: 4952.

Metheall, P., D.C. Barger, RH. Smallwood, and B.H. Brown, 1996, Three-dimensional electrical impedance tomography. Nature 380: 509.

Mintrovich, J., M.E. Mosley, L. Chileuitt, H. Shimizu, Y. Cohen, and P.R Weinstein, 1991, Comparison of diffusion- and T2-weighted MRI for the early detection of cerebral ischemia and reperfusion in rats. Magn. Reson. Med. 18: 39.

Mosley, M.E., Y. Cohen, 1. Mintorovich, L. Chileuitt, H. Shimizu, 1. Kucharczyk, F.M. Wendland, and P.R Weinstein, 1990, Early detection of regional cerebral ischemia in cats: comparison of diffusion and T2-MRI and spectroscopy. Magn. Reson. Med. 14: 330.

Ogawa, S. and T -M. Lee, 1990, Magnetic resonance imaging of blood vessels at high field: in vivo and in vitro measurements and image simulation. Magn. Reson. Med. 16: 9.

Ogawa, S., T-M. Lee, AS. Nayak, and P. Glynn. 1990, Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic field. Magn. Reson. Med. 14: 68.

Ong, K.C., H. Wen, AS. Chesnick, S. Duewell, FA Jaffer, and RS. Balaban, 1995, Radiofrequency shielding of surface coils at 4.0 T. J. Magn. Reson. Imag. 5: 773.

Pennock, J.M., F.W. Cowan, J.E. Schweiso, A Oatridge, MA Rutherfard, L.M.S. Pubowitz, and G.M. Bydder, 1994, Clinical role of diffusion-weighted imaging: neonatal studies. MAGMA 2: 273.

Purcell, E.M., H.C. Torroy, and R V. Pound, 1946, Resonance absorption by nuclear magnetic moments in a solid Phys. Rev. 69: 37.

Rueckert, L., I. Appollonio, J. Grafinaii, P. Jezzard, R Johnson, Jr., D. LeBihan, and R Turner, 1994, MRI functional activation of the left frontal cortex during covert word production. J. Neuroimaging 4: 67.

Stejskal, E.O. and J.E. Tanner, 1965, Spin diffusion measurements: spin-echoes in the presence of time-dependent field gradient J. Chern. Phys. 42: 288.

Stejskal, E.O., 1965, Use of spin echo in pulsed magnetic-field gradient to study anisotropic, restricted diffusion and flow. J. Chern. Phys. 43: 3597.

Stollberger, R and P. Wach, 1996, Imaging of the active BI Field in vivo. Magn. Reson. Med.35: 246.

Turner, R, 1993, Gradient coil design: a review of method. Magn. Reson. Imag. 11: 903. Turner, R, 1991, D. LeBihan, C.T.W. Moonen, D. Despres, and J. Frank, Echo-planar time course MRI of cat brain by oxygenation changes. Magn. Reson. Med. 22: 159.

Ueno, S. and N. Iriguchi, 1998, Impedance magnetic resonance imaging: a method for imaging of

70

impedance distribution based on magnetic resonance imaging. J. Appl. Phys. 83: 6450. Verheul, H.B., R Balazs, J.W. Berkelbach van der Sprenkel, C.A.F. Tulleken, K. Nicolay, K.S.

Tamminga, and M. van Lookeren Campagne, 1994, Comparison of diffusion weighted MRI with changes in cell volume in a mt model ofbmin injury. NMR Biomed 7: 96.

Warach, S., D. Chien, W. Li, M. Ronthal, and RR EdeIm..m, 1992, Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 42: 1717.

Williams, D.S., J.A Detre, J.S. Leigh, and AP. Koretsky, 1992, Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc. Natl. Acad Sci. USA. 89: 212.

Zhang,W., D.S. Williams, and AP. Koretsky, 1993, Measurement ofmt bmin perfusion by NMR using spin labeling of arterial water: in vivo determination of the degree of spin labeling. Magn. Reson. Med. 29: 416.

71

IMAGING BRAIN ELECTRIAL ACTIVITY

Bin He, Dezhong Yao, and Dongsheng Wu

Department of Electrical Engineering and Computer Science and Department of Bioengineering

University of Illinois at Chicago Chicago, IL 60607, USA

INTRODUCTION

Brain activation is a spatio-temporally-distributed process. While the electroencephalogram (EEG) offers excellent temporal resolution to characterize rapidly changing patterns of brain activation, conventional EEG techniques are limited mainly due to their inability to provide spatial information regarding brain activity. As a result, invasive mapping procedures have been increasingly used [Towle et al, 1995]. It is extremely desirable to image spatially distributed brain electrical activity from noninvasive electromagnetic recordings. Such noninvasive means would greatly increase our ability to accurately localize rapidly changing regional patteIns of brain activation, and to aid presurgical evaluation in medically refractory epilepsy.

The poor spatial resolution of EEG is mainly due to the smoothing effect of the head volume conductor. Because the electric potential is generated in the brain and volume conducted from the brain through the low-conductivity skull, and finally to the recording electrodes at the scalp surface, the potential distribution on the scalp is severely smeared and distorted [Plonsey, 1969; Nunez, 1981; Malmivuo & Plonsey, 1995; Gulrajani, 1998]. Such distortion is caused by the basic physical law of the volume conduction, which cannot be overcome by simply increasing the spatial sampling rate.

A number of investigators have attempted to improve the spatial resolution of the EEG. A popular and well-accepted approach, called the dipole localization method (DLM), is a parameter estimation technique, which attempts to account for a scalp potential field by means of one or a few equivalent current dipoles in the brain [Kavanagh et aI, 1978; Sidman et al, 1978; Scherg & Von Cramon, 1985; He et al, 1987; He & Musha, 1992; Homma et al, 1994; Song & Koles, 1995; Fletcher et aI, 1995; Cuffin, 1995; Uutela et al, 1998]. While the DLM has been demonstrated to be useful in locating well localized brain electric events, its utility in imaging spatially distributed brain electric sources is limited because of its

Advances in Electromagnetic Fields in Living Systems, Vol. 3 Edited by J.C. Lin, Kluwer AcademicIPlenum Publishers. 2000 73

fundamental assumption of brain generators. Several methods have been developed in an attempt to image the spatially distributed brain electrical activity without restricting the brain generators to one or few dipoles. The scalp Laplacian method (SLM), which is essentially a high pass spatial filter compensating the distortion caused by the head volume conductor low-pass filtering effect, has been explored due to its capability in enhancing the spatial resolution of the scalp potentials and its reference-free capability [Hjorth, 1975; Perrin et aI, 1987; He & Cohen, 1992; Le et ai, 1992; Nunez et ai, 1994; Babiloni et aI, 1996; He, in press]. The limitation of the SLM is that it provides information on the scalp instead of the brain surface.

Further development has been pursued to achieve brain electric imaging by reconstructing a surface source distribution [Freeman, 1980; Hamalainen & Ilmoniemi, 1984; Nunez, 1987a; Wang et aI, 1992; Srebro et ai, 1997; Baillet & Gamero, 1997; Philips et aI, 1997; He et aI, 1998a], or the cortical potential distribution [Sidman et aI, 1990; Le & Gevins, 1993; Srebro et ai, 1993; Gevins et ai, 1994; Nunez et ai, 1994; He et ai, 1996; Yao, 1996; He et ai, 1997a, b; Babiloni et aI, 1997; Zanow, 1997; Wang & He, 1998; He, 1998; He, in press]. Among these approaches, the cortical-potential imaging has recently received considerable attention. The approach attempts to deconvolve the low-pass spatial smoothing effect of the head volume conductor, thus achieving high resolution imaging over the surface of the brain noninvasively. In addition to these approaches, other techniques have been explored to reconstruct a low-resolution current source distribution inside the brain [PascualMarqui et aI, 1994; Yao & He, 1998]. Multiple dipole distributions have also been explored from noninvasive measurement, using statistical signal processing technology [Mosher et ai, 1992].

The objective of this Chapter is to provide a state-of-the-art review of research efforts on imaging brain electrical activity from the scalp EEG, with focus on model-based spatial deconvolution imaging, especially cortical-potential imaging. We attempt to make the chapter as self-contained as possible. In Section 1, we briefly introduce the concept of volume source and volume conductor. Basic source models and their fields in a volume conductor are discussed in Section 2. In Section 3, we review the inverse imaging algorithms, which are related to EEG spatial deconvolution imaging. In Section 4, we review the recent research activities on EEG deconvolution imaging, with emphasis on corticalpotential imaging in a realistically shaped inhomogeneous head model. Due to limitations in length, it is unavoidable to omit some important research.

1. VOLUME SOURCE AND VOLUME CONDUCTOR

Much is understood theoretically and experimentally about the bioelectric source as volume source and the body as volume conductor [Plonsey, 1969; Plonsey and Barr, 1986; Malmivuo and Plonsey, 1995; Gulrajani, 1998]. Below we briefly summarize the concepts and theoretical treatment of bioelectric sources in a volume conductor.

1.1 Concept of Volume Source and Volume Conductor

As opposed to classic electrical engineering, which deals with networks made up of discrete batteries, resistances, capacitances, and inductors, an important feature of the electrical sources and the conducting medium in living systems is that it is distributed. That is, the conducting medium extends continuously through the three-dimensional volume, serving as a volume conductor. Although the capacitance is localized to cellular membranes,

74

since our interest is at tissue or organ level, the capacitance should also be deemed to be distributed. As discussed below, the inductance of the living tissue as a conducting medium can be ignored for low frequency bioelectric phenomena such as the EEG, evoked potential, and event related potentials. In fact, the distributed nature of the volume conductor is true as well for the bioelectric sources, which are also continuously distributed throughout the excitable medium. Such electric sources distributed in the three dimensional volume is referred to as the volume source [Plonsey, 1969].

1.2 Quasi-static Approximation

In the above description on the living systems as a volume conductor, the capacitive component of tissue impedance is negligible in the frequency band of internal bioelectric events, according to the experimental evidence of Schwan and Kay (1957). Schwan and Kay showed that the volume currents are essentially conduction currents and require only specification of the tissue resistivity. Of importance is the development of a theoretical basis for the quasi-static formulation of bioelectric phenomena, which suggests that time-varying bioelectric currents and voltages in the living system can be examined in the conventional quasi-static limit [Plonsey, 1969]. That is, all currents and fields behave, at any instant, as if they were stationary. The description of the fields resulting from applied current sources is based on the understanding that the medium is resistive only, and that the phase of the time variation can be ignored (i.e. all fields vary synchronously).

1.3 Volume Source in a Homogeneous Volume Conductor

The primary bioelectric sources can be represented as an impressed current density

y (x, y, z, t) [Plonsey, 1969], which is driven by the electrochemical process of excitable cells. In other words, it is a non-conservative current that arises from the bioelectric activity of nerve and muscle cells due to the conversion of energy from chemical to electrical form.

Note that ]; (x, y, z,t) is zero everywhere outside the region of excitable cells [Plonsey, 1969].

If the volume conductor is infinite and homogeneous and the conductivity is a, the

primary sources ];(x,y,z,t) establish an electric field E and a conduction current aE. As a result, the total current density] is given by:

(1.1)

The ohmic current aE is necessary to avoid buildup of charges due to the source current.

Because the electric field E is quasistatic, it can be expressed at each instant of time as the negative gradient of a scalar potential tP, and equation (1.1) may be rewritten as,

(1.2)

According to the quasi-static conditions [Plonsey and Heppner, 1967], the tissue capacitance and inductance are negligible. In other words, the divergence of the total current must be vanishing. Therefore, equation (1.1) reduces to Possion' s equation:

(1.3a)

75

(1.3b)

Equation (1.3) is a partial differential equation satisfied by cP in which v·]; is the source fonction. The solution of equation (1.3) for the scalar function cP for a region that is uniform and infinite in extent [Stratton, 1941; Jackson, 1975] is:

I I -<1> = --JJ-)V ·J'dv

47£0" r (1.4)

Since a source element v·]; dv in equation (1.4) behaves like a point source, in that it sets up a potential field, that varies as I I r, the expression - V .]; can be considered as an equivalent monopole source density.

Using the vector identity v·(J; Ir)=V(l/r)·Y +(I1r)V·]; and the divergence (or Gauss's) theorem, equation (1.4) can be transformed [Stratton, 1 941 ; Jackson, 1975] to:

(1.5)

This equation represents the distribution of potential cP due to the primary bioelectric source ]; within an infinite homogeneous volume conductor with conductivity (}". Here Y dv

behaves like a dipole source (with a field that varies inversely to the power of distance from the dipole source to the observation point). Therefore, the impressed current density]; may be interpreted as an equivalent dipole source density.

1.4 Volume Source in an Inhomogeneous Volume Conductor

Actual distribution of conductivity inside the human body is inhomogeneous. Frequently, a piece-wise homogeneous volume conductor model, (by approximating the inhomogeneous volume conductor by a collection of homogeneous isotropic resistive regions, where the current density Y is linearly related to the electric field intensity E [Schwan and Kay, 1956]) is found to be useful. For such piece-wise homogeneous volume conductor with boundary Sj (where j refers to the j-th homogeneous region), on the

boundaries both the electric potential cP and the normal component of the current density must be continuous [Stratton, 1941; Plonsey 1969]:

<1>' (Sj) = <1>" (S)

O"~V<1>'(Sj)·nj =O";V<1>"(Sj).nj

(1.6)

(1.7)

where the primed and double-primed notations represent the opposite sides of the j-th boundary S j and ii j is directed from the primed region to the double-primed one. If dv is a

volume element, and If/ and are two scalar functions that are mathematically well behaved in each (homogeneous) region, it follows from Green's theorem [Smyth, 1968] that

76

(l.8)

Let IfJ = y,. , where r is the distance from an arbitrary field point to the volume or surface

element in the integration, and <II the electric potential. Substituting equations (1.3), (1.6) and (1.7) into equation (1.8) results in the following [Geselowitz, 1967]:

1 -. 1 1 I . . 1-(r)=-V' ·V(-)dv+-:E .(0". -O"j)V(-)·dSj

41i0" r 41i0" j' J r (1.9)

This equation evaluates the electric potential anywhere within an inhomogeneous volume conductor containing internal volume sources. The first term on the right-hand side of equation (1.9) involving ]1 corresponds exactly to equation (1.5) and thus represents the contribution of the primary source. The effect of inhomogeneities is reflected in the second integral, where (a~ -a)<IIiij is an equivalent double-layer source (iij is in the direction of

dSj ). The double-layer direction, that of iij or dSj , is the outside surface normal (from the

prime to double-prime region). For homogeneous case, (j~ = (j~, thus the second term of equation (1.9) vanishes, and equation (1.9) reduces to equation (1.5).

2. MODELS OF THE VOLUME SOURCES AND THEIR FIELDS IN A CONDUCTIVE MEDIUM

A practical way to investigate the function of living systems is to construct a model that follows the operation of the system as accurately as possible, yet as simple as possible. As the practical observation on brain electric activity is limited by many factors, such as finite spatial sampling, measurement noise, electrode placement uncertainty, etc., a simplified source model is useful in many cases for non-invasive investigation of brain electrical activity. In this section, we discuss the basic elements of the volume source, and present a derivation of the fields produced by basic source elements in a widely used 3-sphere volume conductor model.

2.1 Monopole Source Model

The simplest source configuration is a point source or a monopole. If we consider a point current source of magnitude 10 lying in a uniform conducting medium of infinite extent and conductivity a, then current flow lines must be uniform and directed radially. Consequently, for a concentric spherical surface of arbitrary radius r, the current density J

crossing this surface must be uniform and will equal 10 divided by the total surface area. That is

(2:1)

77

since the total current is conserved. Because the current is everywhere in the radial direction, the current density expressed as a vector is

(2.2)

where 1'0 represents a unit vector in the radial direction, where the origin is at the point source.

Associated with the current flow field defined by equation (2.2) is a scalar potential field <lI. According to the field theory, the electric field E is related to the scalar potential <lI

by

(2.3)

From Ohm's law one has

(2.4)

Collectively, equations (2.2), (2.3), and (2.4) result in

(2.5)

Since the scalar potential is only a function of r, only the component of V<lI in the direction of r can vary. This leads to

10 d<l> --=-CI'-41/7 2 dr

(2.6)

and integration with respect to r leaves us with [Plonsey and Barr, 1986; Malmivuo and Plonsey, 1995]

(2.7)

The potential field described by equation (2.7) for a monopole is identical to the electrostatic potential field from a point charge, provided that 10 is replaced by Qo (the charge magnitude), (1' is replaced by & (the permittivity). Comparing equation (2.7) with equation (1.4) indicates that the source element V·J'dv behaves as a monopole source, and that the electrical potential due to an arbitrary distribution of sources can be obtained by applying the principle of superposition and equation (2.7). The bioelectrical potential expressed by equation (1.4) can also be interpreted as the contribution from a monopole distribution located in a three-dimensional volume space.

78

2.2 Dipole Source Model

In a living system, one can never have a single isolated monopole current source because of electrical neutrality. However, collections of positive and negative monopole sources are physically realizable if the total sum of current is zero. The simplest collection of monopole sources is a dipole, which consists of two monopoles of opposite sign but equal strength 10 (often termed as source and sink) separated by a very small distance, d. In fact, the strict definition requires d ~ 0, I 0 ~ 00 with p = dlo remaining finite in the limit. The quantity p is called the dipole moment or dipole magnitude. In this sense, the dipole model is a more realistic source model than the monopole model, representing bioelectrical sources. Theoretically speaking, the monopole model is the most essential source model, which can generate all other composite source models such as dipole, quadrupole, tripole, and distributed source. Adoption of a specific source model would depend on the specific application.

A dipole is a vector whose direction is defined from the sink to the source

(2.8)

where p is the dipole vector, and do is an unit vector pointing from the sink to source. Fig. 1 illustrates a dipole of arbitrary orientation.

From the discussion in Section 2.1 and the principle of superposition, the electrical potential due to a current dipole source in an infinite homogeneous medium can be obtained by summing the contribution from the source and sink. Suppose the distance from afield point to the source is 1j and to the sink is r2 and to the middle point between the source and sink is r, the distance between source and sink is d, and one half of this distance is represented as d2 (Fig. 1). According to equation (2.7), we have their total potential:

(2.9)

Figure 1. Schematic diagram of a dipole and the illustration for computing the dipole field.

79

According to Fig. 1, we have

d d =r 1+[(-2.)2 -2-2.cosO]

r r

and by Taylor expansion, one obtains

Similarly,

1 d 2 2 d 2 1 d2 2 d 2 0]2 ) r, = r(1 +-[(-) -2-cosO]--[(-) -2-cos + ... 2 r r 8 r r

= r -d2 cosO + d20(d2 ),r» d2 r

r2 = ~r2 + d; - 2d2r cos(l£ - 0)

d = r +d2 cosO + d 20(-2.),r» d 2 r

2 d d So lim r2r1 = r , and r2 - r, = 2d2 cosO + 2dP( -2.) = d cosO + dO( -2.) while according to the

d~O r r strict definition of a dipole requires d ~ 0, I 0 ~ 00 with p = dlo remaining finite in the limit, we have

and finally we get

. pcosO P 1 - 1 1_ <l>d =hm<l>=--=-V(-)·do =-V(-)·p

d~ 41£072 41£(1 r 41£(1 r (2.10)

where the angle () is the polar angle between the dipole vector and the distance vector from the dipole to the observation point.

Comparison of equation (2.10) with equation (2.7) indicates, that the dipole potential field varies as r-2 whereas the monopole potential field varies as r-1• In addition, the dipole equi-potential surfaces are not concentric spheres but rather are more complicated, because of the factor cos() . The maximum dipole potential, for a given value of r, is on the polar axis.

Comparing equation (2.10) with equation (l.5) indicates that the source element Ji dv behaves as a dipole source, and that the electrical potential due to an arbitrary distribution of sources can be obtained by applying the principle of superposition and equation (2.10). Therefore, the bioelectrical potential expressed by equation (1.5) can also be interpreted as the contribution from a dipole distribution located in a three-dimensional volume space.

80

2.3 Other Source Models

In addition to the monopole and dipole models, higher order equivalent source models such as the quadrupole have also been studied to represent the bioelectrical sources [Geselowitz, 1960]. However, the exploration along this line is limited due to the fact that relating the equivalent higher order equivalent source model with physiological phenomena is difficult.

On the other hand, the use of distributed surface source models such as the doublelayer model or the single-layer model have been explored to represent equivalently the volume bioelectric sources. The double-layer model is sometimes also called the dipolelayer model, which consists of a sheet of dipoles over the surface. Similarly, the single-layer model is sometimes called point-source layer or charge layer, which consists of a distribution of point current sources or charges over the surface. The potential field of a double-layer or single-layer source model can be evaluated by equations (l.S) or (1.4), by changing the integral region from the volume to the surface.

As discussed above, the potential fields of an arbitrary source model can be evaluated by applying the principle of superposition and the contribution of a monopole or dipole source. Below we will discuss in detail the potential field of a monopole or dipole in a concentric 3-spheres inhomogeneous model [Rush and Driscoll, 1969]. Such a volume conductor has been widely used to represent the head volume conductor, or to validate numerical algorithm for fields in an inhomogenous head model of arbitrary shape.

2.4 Electrical Potentials in a Concentric 3-Spheres Conductor Model

Among various head volume conductor models, the concentric three-sphere model [Rush and Driscoll, 1969] has been widely used to represent the head volume conductor [Kavanagh et aI, 1978; Nunez, 1980; Nunez, 1987; He et aI, 1996; Wang & He, 1998; He et aI, 1998a; Yao & He, 1999]. In the 3-sphere model, the inner and outer radii of the skull are usually chosen to be 8 and 8.5 em, respectively, while the radius of the head is 9.2 em. For the brain and the scalp, a typical resistivity of 2.22Qm is selected, whereas for the skull a resistivity of 80 x 2.22Qm = 177Qm is assumed. These numerical values are given solely to indicate typical resistivity (conductivity) values corresponding to each region inside the head. The concentric 3-sphere head model has been used widely in the brain forward and inverse problems. Below we present derivation of the potentials on the scalp and inside the brain by a dipole or a point current source.

The potential distribution of an arbitrarily positioned and orientated dipole in an infinite medium is given by the following [Stratton, 1941]:

<1>. =-p·v (-)=- P -+P ---+P -- -I - I I{ a I a I a}I E 4JZ"CT 0, rp 4JZ"CT '&0 'I' ro sin Bo arpo 6 ro aBo rp

(2.11 )

For a point current source in an infinite medium, the potential is given by:

(2.11a)

81

where, I is the current source strength. Using the associated Legendre function, we can obtain [Hobson, 1931],

(2.l2)

(2.13)

and

(2.14)

where &mo = 1 when m = 0, and &mo = ° when m "* 0, and (ro,f)o'rpo) and (r,{),rp)are the standard spherical coordinates of the source position and field point, rp is the distance from

the source point to the field point, P,~ is the associated Legendre function of the first kind. From equations (2.l2)-(2.l4), equation (2.11) can be rewritten as

where

=~ ~K,m(p) sm(p () ) LL- 1+1 I "qJ 1=1 m=O r

mP H; = IP,p,m (cos{)o)cos mrpo __ ._'I'_p'm (cosOo)sin mrpo -

sm{)o

Po «(1- m + 1)(1 + m)p,m-I (cos{)o) - p,m+1 (cos 00 »cos mrpo 2

mP G,m = IP,P,m(cos{)o)sinmrpo +-._'I'_p,m(cos{)o)cosmrpo

sm{)o

Po «(1- m + 1)(1 + m)p,m-I (cos{)o) - p,m+1 (cos{)o »sin mrpo 2

and equation (2.l1a) changes to

cI> 00 = t ± K,~~/) (cosmrpo cosmrp + sin mrpo sin mrp)p'm (cos{)o)p'm (cos{) f=lm"'O r

= ~ ~ K,m(I) sm(J () ) L £.. 1+1 I "lP 1=1 m=O r

82

(2.l5)

(2.15a)

(2.15b)

(2.l6)

where

S;(l,8,rp) = (cosmrpo cosmrp + sinmrpo sinmrp)p'm(cos8o)lncos8)

K;(l)= Ir; (2_0 o)(/-m)! 41iCT m (I + m)!

(2. 16a)

(2.16b)

Note that the index 1 starts from 1 in equation (2.15) instead of 0 as in equation (2.12), since the term corresponding to 1 = 0 is zero. In equation (2.16), the 1 = 0 term was omitted due to the electrical neutrality for a living system that requires the total sum of the currents inside a living system vanishes.

In the above equations, S;(P,(},rp) is decided by the dipole position and moment, and

has no relation to the radial variable r, K;(P) is a constant which is totally determined by the

dipole, Sr{l,(},rp) is decided by the point current source position and its strength, and has no

relation to the radial variable r, and K;(l) is a constant which is totally determined by the point current source. Because of the similarity between (2.15) and (2.16), a general formula is given below, in order to simplify the following derivation.

<1>., = L ~~ S;(8,rp) I,mr

(2.17)

For a dipole, K; =K;(ft), S,m (8,rp)=S;(p,(},rp); while for a point current source,

Kim = K; (l) and S,m «(},rp) = S,m (l,(},rp). We will not specify them in the following derivation for the sake of simplicity.

As the boundary condition requires that the potential is continuous at the boundary, which is shown by the following equations (2.21) and (2.23), so, for a concentric multisphere conductive medium S;«(},rp) may be chosen as the common basic function form for each sub-region. Based on equation (2.15) and (2.16), the expression of the potential in the inner sphere can be assumed to be the following

<1>, = :E[A;r l + K; -b-1,; I,m r r (2.18)

In the second and third layers the potentials are expressed as the solution of Laplace's equation:

(2.19)

(2.20)

The boundary conditions are:

(2.21)

83

(J' act> 1 = (J' act> 2 1 1 ar 2 ar ,."

<1>3 = <1>2\ r=h

8<1> 3 8<1> 21 (J'3&=(J'2& ,=b

a:;3 =Ol,=c

(2.22)

(2.23)

(2.24)

(2.25)

In the above, (J'1,(J'2 and (J'3 refer to the conductivies for the brain tissue, the skull and the scalp, respectively. a, b and c refer to the radii of the three spheres. The boundary conditon (2.21) implies:

em 1 D'" 1 A'"' Km 1 ,a + , ---,,:J= ,a + ,-,-, a a+

(2.26)

The boundary condition (2.25) yields:

E mlc'-I_F.m (l+I)_I_=O , , C'+2

(2.27)

or

(2.28)

From the boundary condition (2.23) we have

(2.29)

and the boundary condition (2.24) results in

(J' [e'"lb'-I - D"(l + 1)_1_] = (J' F.'"(/ + 1)[b'-lc -(2/+') __ 1_] 2' , b'+2 3' b'+2 (2.30)

Equating F,m from the above two equations we find

e,mb 2l+' + D,m

( b)21+1 1+1 - -+1 c I

em _1_b 2l+1 _ D'" (J'2 ' 1+1 '

(b)21+1 - -1 c

(2.31)

or

84

(2.32)

which we abbreviate as:

(2.33)

The boundary condition (2.22) yields

0" [Am la'-' - K m (I + 1)_1_] = 0" em [Ia'-' - (I + 1) a a-(1+2)] 1 / I a l+2 2 I P (2.34)

and the boundary conditon (2.21) is

Alii a ' + K m _1_ = em [a' + a a-(I+I)] 1 1 a '+1 1 f3 (2.35)

By equating A;" from the above two equations we find

c'(' = K'(' (21 + 1)

1(1- 0"2 )a 21+1 + ~(l + 0"2 (l + 1)) 0"1 jJ 0"1

(2.36)

and

Am -Km( (2/+ l)(jJ + aa-(21+I») I - I a-(21+1»)

1(1- 0"2 )f3a21+1 +a(1 + 0"2 (I + 1)) (2.37)

0"1 0"1

Collectively, equations (2.28)-(2.29), (2.33), and (2.36) result in

(2.38)

Based on this equation and equation (2.20), we can calculate the potential on the scalp surface by

<P I = I[Em ( 1+_1_ 21+1 -(1+1») l(,m 3 '=c I.m leI + 1 C C r I

=IEmc,2/+1sm I.m I 1+1 I

(2.39)

85

And the potential in the brain (To <r ~ a) can be calculated by equations (2.15)-(2.18) and equation (2.37). In particular, the potential on the cortical surface (r=a) is

(2.40)

Considering a special case where 0"1 = 0"3 and the source is located on the z-axis with z = To,

the potential on the scalp produced by a point current source located on the z-axis can be represented by

(2.41)

and the potential on the scalp produced by the z-component Pz of a dipole located on the zaxis is [Eshel, 1993]

where

86

 I = ~ P)r:-I s(21 + 1 )3 Po ( 0) 3r=c £.. II I cos

1=' 41f0"]c + dl(1 + 1)1

= f: P.lf l-

I s(21 + 1 )3 Po (cosO) 1=1 41f0"3 C2 dl(1 + 1)1 I

d l = «I + 1)s+/)(~+ I)+(l-s)((l + 1)s+/)(j;" - I;' )-/(I-s)2(IJj )", 1+1 Ih

II =2/+1,

it =%, 12 = 'ie, I='i.

(2.42)

The potential on the cortical surface (r=a) produced by a point current source is

and the potential on the cortical surface (Fa) produced by a dipole is [Eshel, 1993]

where

c{ = (2/ + I)(f/ + ; J,-({+I»,

d{ =1(1-s)J,2{+1 + a (l+(1+I)s), fJ

a = f22{+1 (1- s) - (1 + I ~ I)'

fJ = f2-(2{+I)(1-S)_(l + / + Is). I

(a)

(b)

(c)

(d)

(2.43)

(2.44)

Figure 2. An example of scalp potential map (a) and cortical potential map (b), generated by two radial dipoles located at (0.6cos(1t/6), ±0.6sin(1t/6), 0) in a concentric 3-sphere inhomogeneous head model. The scalp potential map (c) and cortical potential map (d), generated by two point current sources located at (0.7cos(1t/6), ±0.7sin(1t/6), 0) and two point current sinks located at (0.Scos(1t/6), ±0.Ssin(1t/6), 0) in the same spherical model, are also shown. Note the enhanced resolution of the cortical potentials as compared to the scalp potentials.

87

Fig. 2 shows an example of the potentials on the scalp (a and c) and the cortical surface (b and d), by two dipoles (a-b) and 4 current point sources/sinks (c-d), in the concentric 3-sphere head model. Note that the cortical potential provides enhanced resolution in identifying and resolving the actual sources, as compared to the scalp potentials. It is also of interest to note that, the two pairs of source-sink (1.84 cm apart) source configuration present an indistinguishable potential distribution on both the scalp and cortical surface, as compared to the potential distribution produced by the two dipoles.

3. Algorithms for EEG Deconvolution Imaging

3.1 Concept ofEEG Spatial Deconvolution

In the previous section, we discussed source models and the forward solutions, i.e., calculation of the potential field by a given source model in a volume conductor model. In practice, it is important and interesting to infer the origins from the scalp-recorded EEG, or to localize the sources that generate the observed EEG on the scalp. Such a problem is called the inverse problem. The inverse problem has no unique solution, as the potential distribution measured on the outer surface of a volume conductor can be produced by an infinite number of sources within the conductor [Helmholtz, 1853].

The bioelectric inverse problem, which shall be defined as the problem to reconstruct source distributions from body surface electrical measurements, is intrinsically underdetermined. The source localization method does not lead to a unique solution of the inverse problem unless further constraints are considered, for example, the sources are restricted to the cortex, or some physiological constraints and mathematical regularization techniques are included. For the various spatial deconvolution imaging techniques that are discussed in this chapter, the brain electric sources are generally represented by an equivalent surface source or a direct reconstruction of the cortical surface potential by transfer matrix method. The common mathematical model among all these spatial deconvolution imaging approaches is a linear system, that links the equivalent surface source parameters (or cortical potentials) to the potential measurements on the scalp via a transfer matrix, which is physically unique. This mathematical model can be written as

D=GX (3.1)

where D is the vector consisting of scalp-recorded potentials, X is the unknown vector, and G is the transfer matrix. The detailed form of X and G will be described in the next section with regard to specific models. The inverse problem of interest to us is to seek X from D. As the measurement electrode number is always much smaller than the dimension of the unknown vector X, this problem is an underdetermined non-unique inverse problem. A proper regularization strategy is necessary for obtaining a reasonable solution to this problem.

3.2 General Inverse and SVD Algorithm

3.2.1 General Inverse

Since noise always exists in scalp measurements, equation (3.1) should be modified to

88

D=GX+n (3.2)

where n is a noise vector whose dimension is the same as the scalp recording data. Based on equation (3.2), a least-square estimate can be obtained by minimizing the following objective function

O(X)=IID-GXI1 2 ~min

The following equation is obtained from equation (3.3)

If 0 is a nonsingular square matrix, we have

and for an arbitrary matrix, we have

X = (G TGfGl'D

=GT(GGTfD=G+D

(3.3)

(3.4)

(3.5)

(3.6)

where 0+ signifies the general inverse. The general inverse solution is a minimum norm (MN) solution [Golub et al, 1989] among the infinite set of solutions which satisfy the scalp measurement potentials, and has been used in MEG and EEG inverse solutions [Hamalainen & Ilmoniemi, 1984, 1994; Crowley et aI, 1989; Sidman et al, 1990; Wang et al, 1992].

If (OTO) is invertible, we have

(3.6a)

This case corresponds to the overdetermined situation which means that there are more independent measurements than independent unknown variables, and 0 is noted as a nonsingular column-matrix.

If (OOT) is invertible, we have

(3.6b)

This case corresponds to the underdetermined situation which means that there are less independent measurements than independent unknown variables, and G is noted as a nonsingular row-matrix.

Since the matrix inversion in equations (3.5), (3.6a), or (3.6b) is simple and can be considered as a special case of the general inverse, we will only investigate the algorithm for equation (3.6). Also, only the singular value decomposition (SVD) algorithm will be introduced to compute the general inverse, because it is the most popular one in the engineering field and its detailed algorithm and software can be found in many textbooks [Press et aI, 1992].

89

3.2.2 SVD Algorithm

For an m x n matrix G, we have its SVD as [Golub et al, 1989]

G=ULV' (3.7)

where

The vectors U; and V; are the ith left and right singular vectors of dimension m x 1 and n xl,

respectively. The.4.; are the singular values of matrix G, L is a diagonal matrix with the singular values on its main diagonal. Furthermore, ifrank(G)=r, i.e.,

(3.8)

then the SVD expansion is defined by

(3.9)

where u, and v, are m x r and n x r matrices composed of the related r singular vectors, and

L, is the r x r diagonal matrix with the first r singular values on the main diagonal. Based on the SVD of matrix G, the Moore-Penrose inverse or "pseudo-inverse" of G

is defined as

(3.10)

For the linear system of equation (3.1), the solution X estimated by equation (3.6) is

(3.11)

3.3 Weighted General Inverse Method

The above general method is based on the fact that the general inverse of the matrix G can be correctly calculated, and the only problem is a proper choice for truncation. However, because the transfer matrix for spatial deconvolution imaging is ill-posed, the calculation of the general inverse is not an easy problem. Therefore, a proper modification to the matrix before the inverse calculation is important to obtain a good inverse. A few methods have been proposed to modify the transfer matrix, and are described below.

90

3.3.1 Measurement Space Weighted MN Solution

Based on equation (3.3), the MN solution is also the least square estimate of the L2 norm objective function of the residue errors [Gencer et al, 1998]. The larger the residue error for an electrode, the larger weight the related measurement electrode has. In general, a left weighting factor Wd may be incorporated into the algorithm to emphasize the importance of certain electrodes over others:

(3.12)

or

(3.12a)

and

(3.12b)

There are at least three examples for the measurement space weighted MN algorithm:

1) Equation (3.4), where Gl'D = Gl'GX,Wd = Gl';

2) Laplacian weighted algorithm, where

L,D = L,GX (3.13)

Wd = Ls is a Laplacian operator acting on the measurement space [He, 1994; Wu et al, 1995; He & Wu, 1997]. The Laplacian weight makes the least square fit to pay more attention to the high spatial frequency component in the measurements, thus compensating the high spatial frequency components lost during the volume conduction.

3) Noise convariance weighted algorithm: In the objective function defined by (3.3), an equal weight is given to each measurement. The convariance is incorporated into the objective function to balance the weight of each measurement electrode, i.e.,

O(X) = (D -GXl C(D -GX) ~ min (3.14)

and it results in

(3.15)

91

By comparing equation (3.15) with equation (3.12b), one can see that W;Wd = C- = Cov(nr . Recent applications in the EEG inverse problem can be found in Van Veen et al (1997).

3.3.2 Solution Space Weighted MN Solution

The above weighting factors try to modify the "weighting" effect of the different measurement electrodes. There is another kind of problem, which is associated with the minimum norm characteristic of the solution. For example, a number of numerical experiments show that the MN solution tends to be at a more superficial depth than the actual source depth. It is because for the same scalp potential distribution, the column norm of the transfer matrix from a shallow position is larger than that from a deeper position, thus making the norm of the solution smaller for the shallow position than that from the deeper position corresponding to the same scalp potential distribution. So the shallow position is preferable to the deep position in the sense of minimum norm requirement in the MN solution. In order to overcome such probems, another weighting factor should be introduced to balance the relative "weighting" of different position in the solution space. This can be achieved by modifying equation (3.1) as follows

D=GW;WxX=GrX' (3.16)

where Wr is a weighting factor acting on the solution space,Gx = GWr- ,X'= WrX' and then we have

x = Wr- X'= Wx-G;D (3.17)

Some weighting factors used in brain inverse problems are listed below as examples.

1) Normalized Weighted MN Solution: In this approach, in order to balance the effect of the difference between the norm of

each column of the transfer matrix, the norm of each column of the transfer matrix is used as the weighting factor for the corresponding position in the solution space [Jeffs et ai, 1987]. In this way, the norm of the transfer matrix is normalized,

(3.18)

2) Laplacian Weighted MN Solution: Suppose we have a weighting factor like

(3.19)

where L is a Laplacian operator acting on the solution space. For a three-dimensional solution space, it is a three-dimensional discrete Laplacian operator, and for a twodimensional solution space, for example on the cortical surface, it is a two-dimensional

92

Laplacian operator. This approach changes the minimum norm to the minimum norm of the high frequeny component [pascual-Marqui et al, 1994].

3) Local Average MN Solution:

In the cases where no additional a priori information about the sources is availabe, a natural constraint posed by the underdetermined character of the EEG inverse problem is the estimation of averages of the current density at each solution point [Grave de peralta Menendez et aI, 1998]. Suppose,

X"",,=AX (3.20)

where A is a local average operator, and so WX=A. Different kinds of average can be constructed, and the inverse solution will depend on the specific form of A. For example, an element of matrix A may be defined as

A = -dJ 1K ij e ; (3.20a)

where d ij represents the Euclidean distances between the ith and jth grid points. The

coefficient Ki is a rectification coefficient. It is of interest to note that the weighting factors for the measurement space

represented by equations (3.13) and (3.15) are designed to make a more efficient utilization of the measurements; while the above weighting factors for the solution space represented by equations (3.18), (3.19) and (3.20a) are designed to compensate for the superficially shift of the MN solution, or in a more general sense, the bias of the solution. However, another shortcoming of the MN solution has not be considered, i.e., the MN solution is an oversmoothed solution of the actual localized neural electrical activity. In order to enhacne the spatial resolution of the MN solution, recursive weight MN solution has been developed, such as:

(3.21)

In this approach, the previous k-Ith solution X k _1 is recursively used in the kth iterative calculation as a weighting factor [Gorodnitsky et al, 1995], where k = 1,2, ...... , and Xo is the normalized MN solution shown by equations (3.17)-(3.18). As the weighting factor shown by (3.21) will enhance the stronger components in the solution space, step by step, in a sense of weighted MN, it will eventually result in a higher resolution than the initial MN solutionXo. Certainly X k _1 may be replaced by a function of X k _I'XH , •• xO •

Apparently, we can construct similar weighting factors, such as the 1st or the 3rd derivative of the solution, and various combination of the above weighting factors. Of importance is to seek a proper weighting factor or a proper combination of weight factors for a specific problem.

93

3.4 Constrained Inverse Method

In the above section, the main effort is to modify the transfer matrix, and similarly, we can also try to introduce some constraints to the objective function by similar considerations listed above.

In general, the objective function may be considered to be a trade-off between a few related properties such as

(3.22)

Minimizing the above objective function results in the following equation

(3.23)

where the term related to fixllwxxI12 may be some spatial constraint on the solution, and Wx may be the various weighting factors listed above or other factors. fi,IIw,xI12 may be some temporal constraint on the solution. Two methods oftern used to incorporate the temporal constraint are described below.

1) Twomey Regularization Method [Twomey, 1963]: In this regularization strategy, instead of imposing constraints on the magnitude of the

solution or on its function, this method minimizes the difference between the solution, X, and some initial estimate of the solution, X 0' which may be the estimate at the last temporal point. Supposing fix = o.o,w, = I, then equation (3.23) may be changed to

and it can be rearranged as

and so we have

(X -Xo)=[GTG+P,IJGT(D-GXo)

X =(GTG+ P,If(GTD+ P,Xo)

Equation (3.26) has been used in the ECG inverse problem [Oster & Rudy, 1992].

2) Orthogonal Projection Temporal Contraint Method (Baillet & Garnero, 1997):

(3.24)

(3.25)

(3.26)

This method use X i-I as the unique temporal neighbour of X. Assume that X k-I and X may be very close to each other, that is, the orthogonal projection of X on the hyperplane E:L, perpendicular to X k_1 is "small." Thus, we get a weight as

94

(3.27)

where I is an identity matrix. In general, we may consider even more terms in equation (3.22) to get a more general

formula. However, the more terms mean more unknown parameters f3 to determine. The commonly used strategy is to keep one constraint term only. So, equation (3.23) becomes

(3.28)

and

(3.29)

Equation (3.29) is usually called the Tikhonov regularization scheme [Tikhonov, 1977], where f3 is the regularization parameter. When R = I (the identity matrix), Eq. (3.29) corresponds to the Tikhonov zero-order regularization. When R = G (the gradient operator), Eq. (3.29) corresponds to the Tikhonov first-order regularization. When R = L (the Laplacian operator), Eq. (3.29) corresponds to the Tikhonov second-order regularization. Extending to this standard concept, the above listed operators can be used as R, too. Examples of the use of Tikhonov regularization schemes in the EEG inverse imaging can be found in [Srebro et ai, 1993; Lian et ai, 1998; He et al, 1998a].

3.5 Regularization Method

For the weighted minimum norm solution, we need to decide the level of SVD truncation, and for the constraint inverse method, we need to determine the regularization parameter f3. Both problems are usually called regularization.

3.5.1 Truncated SVD Pseudo-Inverse

If the noise n in measurements is additive, the MN general inverse solution will be [Shim et ai, 1981; Sullivan et ai, 1984]

(3.30)

In such case, the condition number of the matrix G (the ratio of maximum to minimum singular value) and the noise level in the measurements become important. The second term of equation (3.30) may dominate the first term, because of small singular values. The noise level in the measurements must be sufficiently small to compensate for the effects of the smallest singular values. In cases where this is not possible, the effects of the second term can be eliminated by using the method of the truncated pseudo-inverse method [Shim et ai, 1981]. This method is simply carried out by truncating at an earlier index p<r in the evaluation of G+ given by equation (3.6). However, while this method reduces the effects of measurement noise on the reconstructed inverse solution, it also imposes a severe loss of

95

information because of the exclusion of small singular values. The optimal truncation index can be found as [Shim et al, 1981; Jeffs et al, 1987; Gencer et ai, 1996]

= {k A! > ECllnI12)) Popl mttx I A~ - E(lln I12) (3.31)

Assuming X and n are independent white Gaussian vectors, the operators EO and IHI represent the expected value and Euclidean norm, respectively.

3.5.2 L-curve Method

For the constraint inverse equation (3.24), there is no proper method to decide two or more regularization parameters, so they are usually decided by some empirical method with respect to a special problem. For the most popular equation (3.29), there are several wellknown methods to determine the regularization parameter. All of these approaches rely on a trade off between the norm of the residual, liD - GXIl ' and the norm of the solution, IIxli.

Hansen (1990, 1992) popularized the L-curve approach to determine a regularization parameter for the ill-posed inverse problem, which was first described by Miller (1970) and by Lawson and Hanson (1974). The L-curve approach involves a plot, using a log-log scale, of the norm of the solution, IIxll, on the ordinate against the norm of the residual, liD - GXIl '

on the abscissa, with f3 as a parameter along the resulting curve. In most cases the shape of this curve is in the form of an "L", and the f3 value at the comer of the "L" is taken as the result. At the comer, clearly both liD - GXIl and IIxll attain simultaneous individual minima that intuitively suggests an optimal solution. Hansen (1990, 1992) also showed that this comer will exist as long as three conditions are met. The first is the discrete Picard condition. This is simply the condition that, on the average, the coefficients w(i) decay to zero faster than the nonzero singular values A.i for the unperturbed problem, in other words, with no geometry error in G and no measurement error in D. The discrete Picard condition may be assumed to be satisfied for the bioelectric inverse problems. Hansen's second condition is that the noise vector n in D is a random vector of zero mean and with a covariance matrix proportional to the identity matrix. This requirement can be presumed to be approximately fulfilled if the uncorrelated Gaussian measurement noise in D dominates the more highly correlated geometry error in G. The final condition is that IInll < liD - nil, i.e., the norm of the noise vector is smaller than the norm of the exact noise-free measuremnt. This condition is needed simply to ensure an adequate signal-to-noise ratio.

A numerical algorithm to compute the site of the L-curve comer, when it exists, has been given by Hansen and O'Leary (1993). The algorithm is based on identifying the comer, and therefore PL. Recently, Lian et al (1998) suggested a minimum product method to determine the L-curve comer. In the minimum product method, assuming the existence of "L" curve has been guaranteed by the model, the minimum area corresponding to the rectangular region determined by the origin and a point on the curve should correspond to the comer of the curve, and so the point corresponding to the minimum product of the norms of the residual and the solution can be used to determine the regularization parameter. The Lcurve method can also be used to determine the truncation level in the truncated SVD method [Hansen, 1990].

96

3.5.3 CRESO Algorithm

Colli-Franzone et al (1985) proposed an empirical approach for determining the regularization parameter f3, denoted composite residual and smoothing operator (CRESO). In this approach, the f3 is determined as the smallest value of P > 0 that results in a relative maximum of the function

(3.32)

Based on the singular value decomposition of G, and letting R = I, this function may be changed to

(3.33)

Apparently, using equation (3.33) is far more accurate and convenient than using a numerical differentiation to compute equation (3.32). As the regularization parameter f3 is explicitly used in the algorithm, this approach can only be used for constrainted inverse methods and not for truncated SVD methods. Our recent experience suggests the feasibility of applying the CRESO algorithm to cortical imaging [Lian et al, 1998; He et al, 1998a].

3.5.4 Discrepancy Technique

Based on the discrepancy principle [Morozov, 1967, 1984; Kirsch, 1996], we compute f3 such that the corresponding Tikhonov solution X, i.e., the solution of the equation

PRJ( +GTGX =G7'D (3.34)

that satisfies the equation

IIGX - DII = Ilnll = E (3.35)

Our recent experience indicates that, as a regularization method, the discrepancy principle is applicable to the EEG inverse problem (Yao & He, 1998).

4. EEG Inverse Imaging of Brain Electric Activity

4.1 Introduction

Among many EEG inverse solutions, the spatial enhancement approaches offer improved spatial resolution as compared to the conventional scalp EEG, meanwhile no ad hoc assumption is made on the underlying brain electrical sources. This is especially true comparing with the current success of the dipole localization method, which can localize in

97

3-dimension a well-localized brain electrical source. The unique feature of the spatial enhancement approach is its applicability to all kinds of brain electrical sources, thus having the potential of achieving our ultimate goal of imaging brain electric activity.

Of particular interest is the recent development of cortical imaging approaches, in which an explicit biophysical model of the passive conducting properties of the head is used to deconvolve a measured scalp potential distribution into a distribution of electrical potential or current source on the cortical surface. Because the cortical-potential distribution can be experimentally measured and compared to the inverse imaging results, the cortical-potential imaging approach provides a powerful imaging means, which is also of physiologic importance. Furthermore, the cortical potentials offer a much-enhanced spatial resolution in assessing the underlying brain activity as compared to the smeared scalp potentials.

Several kinds of EEG inverse imaging approaches have been explored. One approach is to image a source distribution from the scalp potentials in the source space [Freeman, 1980; Hamalainen & Ilmoniemi, 1984; Nunez, 1987a; Wang et aI, 1992; Srebro et al, 1997; Baillet & Garnero, 1997; Philips et aI, 1997; He et al, 1998a].

The brain deblurring method, reported by Gevins and co-workers [Le & Gevins, 1993; Gevins et aI, 1994], estimates potentials at the superficial cerebral cortical surface from EEG recordings on the scalp using a finite element model of each subject's scalp, skull and cortical surface constructed from their magnetic resonance images. In this method, Poisson's equation is applied to a conducting volume between the scalp and the cortical surface, and finite element method is used to handle the complex geometry and varying conductivity of the head. Reported predictions of cortical potentials are quite accurate in the cases shown and dramatic improvement in spatial resolution is achieved.

Srebro e/ al (1993) linked the evoked potential field on the scalp with the brain surface field by Green's second identity. The volume conductivity between the surfaces is assumed to be homogeneous and detailed anatomical information for each subject is obtained from MR images. Regularized inversion is applied to get the cortical surface potential estimation. Their physical and human experiments demonstrate that the estimated epicortical potential fields are more focused than their scalp field counterparts and could also provide useful information for localizing cortical activity from visual evoked potential (VEP) scalp fields.

He et al.(1997a, b, 1998b) recently developed a boundary element method (BEM) based imaging technique, in which both the realistic geometry and the inhomogeneity of the head can be taken into account. The head is represented by 3-shell realistically shaped conductors, each representing the scalp, the skull, and the brain tissue. The detailed anatomical information is obtained from the subject's MRI. Cortical potentials are inversely reconstructed from the scalp potentials via a regularization algorithm. Computer simulation studies and initial human studies demonstrate the feasibility of this BEM-based cortical imaging methodology.

Sidman et al. used a hemisphere equivalent dipole layer, to generate an inward harmonic potential function in a homogeneous sphere head volume conductor model, and then reconstruct the potential at an image surface, including the cortical surface [Sidman et aI, 1990, 1992]. A similar implementation has also been reported using a hemisphere charge layer in a homogeneous sphere [Yao, 1996]. Recently, several algorithms which take into account the skull inhomogeneity were reported, where cortical potentials are reconstructed from the general equivalent source distribution [He et al, 1996; Babiloni et al, 1997; Zanow, 1997; Wang & He, 1998].

Below, we discuss representative imaging approaches as applied to cortical-potential imaging. The categorization is chosen for the purpose of clear presentation of these

98

important research activities, and emphasis is given to those methods which can lead to an inverse imaging solution in a realistically shaped inhomogeneous head model.

4.2 Cortical-Potential Imaging via Equivalent Sources

4.2.1 Principles of CIT Algorithm

The cortical imaging technique (CIT) algorithm, proposed by Sidman et al (1990), has played an important role in the development of EEG inverse imaging methodologies. Because its principles are similar to several methods developed recently to achieve corticalpotential imaging, Sidman's CIT algorithm will be discussed briefly below.

Let the head be simulated by a homogeneous sphere of radius 1. There are two hemisphere shells within as well as concentric.to the scalp. The first hemisphere represents an image surface (which includes the cortical surface) where the potential field is reconstructed. The other hemisphere is a hypothetical dipole layer where dipoles are reconstructed to generate the measured scalp potential as precisely as possible.

Assume there are N radial dipoles, D; (i = 1, ... , N), evenly distributed over the dipole layer. The location and orientation of each dipole are fixed. Only the strengths of the dipole moment, ii, are unknown variables. From scalp electrodes, A; (j = 1, ... , M), the potential measurement Vj can be obtained. The relation between the scalp potential measurement and the dipole layer can be constructed by a system oflinear equations:

N

~.t;*Uji=Vj ;=1

(4.1)

where j = 1, ... , M, and Uj; is the value of potential at the electrode position Aj generated by the dipole D; of unit moment strength. Equation (4.1) can also be presented in a matrix format:

(4.2)

where U is a M*N matrix consisting of Uj;. F is the column vector of dipole moment strengths, and V is the vector of scalp measured potentials.

As opposed to early approaches in spatial deconvolution from scalp potentials, Sidman et al used more dipoles (160) than the scalp recording electrode (33). Thus, equation (4.1) is an underdetermined system because M < N, which leads to infinite numbers of solutions. To obtain a unique solution, the minimum norm algorithm [Hamalainen and Ilmoniemi, 1984] is applied. The truncated SVD algorithm discussed in section 3 was used to obtain a unique solution for equation (4.1) of minimum norm. The truncation level was empirically chosen and reasonable experimental results were obtained [Sidman et al, 1992]. Once the dipole moment strengths are determined, the potential distribution on an image surface is calculated directly through forward solution. While the "image" surface in Sidman's CIT method includes the cortical surface, the most effective depth for the dipole layer in Sidman's implementation is at the radius of 0.45 [Sidman et al, 1992]. Numerical experiments have shown that extending the dipole layer to a more eccentric level close to the cortex will introduce numerical errors, which are caused by the use of the homogeneous head model. Several methods have been reported recently to achieve the cortical-potential imaging, which is essentially based on the CIT's Principles.

99

4.2.2 CIT in a Concentric 3-Sphere Inhomogeneous Head Model

In this section, we discuss a new CIT algorithm, which we have recently developed in our laboratory at the University of Illinois at Chicago [He et al, 1996; Wang & He, 1998]. The motivation for our work and the improvements we have made are described below.

The mathematical principle of CIT is that it is a solution of the inward harmonic continuation problem. Values of a harmonic function (the potential) are measured on the boundary of a volume conductor (the head). CIT continues this function into the interior of the conductor (the cortical surface) by constructing a dipole layer in the brain. From the point of view of electromagnetic theory, a closed surface of dipole layer is preferable to equivalently representing the sources enclosed by this surface. By constructing and determining a closed dipole layer, the potential everywhere between the scalp and this closed surface can be reconstructed theoretically under the condition previously mentioned [Yamashita, 1982]. Therefore, our first improvement is the replacement of the hemispherical dipole layer with a spherical dipole layer.

The second improvement we have made is to adopt the concentric 3-spheres inhomogeneous model to represent the head volume conductor. In Sidman's implementation, the head was taken as a homogeneous sphere with a radius of unity. Such a model does not take the variation in conductivity of different tissues into consideration. The lack of skull low conductivity inhomogeneity causes significant error in cortical imaging of cortical sources. Fig. 3 illustrates our CIT model. The normalized values of the conductivity of the scalp and the brain were taken as a = 1.0, and that of the skull as as = 0.0125. The transfer matrix in equation (4.2) can be evaluated by equation (2.41).

CORTICALSUJlPACB SX1ILL

DIJIOlB LAYER

Figure 3. Schematic illustration of our source-conductor model for the spherical CIT (From Wang and He, 1998. © 1998 IEEE).

In order to use CIT and spherical dipole layer to perform cortical-potential imaging of superficial sources, the dipole layer should be superficial enough to confine these sources. This means that the radius of the dipole layer sphere should be increased to the range where the cortex lies. With the extension of this surface, the number of dipoles on the surface should also be increased to keep the dipole layer dense enough to represent the sources

100

within. Instead of using 160 dipoles adopted by Sidman et al., we use a much larger number of the dipoles and determine the suitable number of the dipoles through computer simulation for a specific dipole layer surface.

Finally, as the dipole layer is extended to the more superficial position, higher spatial frequency components are generated at the scalp, which require a high rate of spatial sampling. A large number of electrodes is required for both the cortical sources and the superficial dipole layer. In the previous implementation of CIT [Sidman et al, 1990], 28 or 36 scalp electrodes were employed. We have used a large number of electrodes (128 electrodes) in our CIT procedure.

After the dipole layer strength vector F was determined, the potential distribution over the cortical surface can be constructed from this dipole layer by using equation (2.43).

.. Q .. ..

[;oil ~

.~ ... ~ ~

0.8

0.6

0.4

0.2

o.o+--~~~--~-----,.----~---~--~ 0.2 0.3 0.4 O.S

Eccentricity

0.6 0.7 0.8

Figure 4. Effect ofthe number of the dipoles to the CIT results when 10% Gaussian white noise is added to the scalp potentials. Three-sphere head model, 128 electrodes, a spherical dipole layer at a radius ofO.75, and the optimal noise ratio were used in the simulation. Different dipole numbers--31 0, 1292 and 5182, were tested, corresponding to different curves in the figure. Two radial dipoles, located at (±r' sin(7Z" /6),0,r' cos(7Z" / 6)) were used as the source. Notice that substantial improvement was achieved when the dipole number was increased from 310 to 1292, whereas no obvious improvement can be achieved when the dipole number was increased from 1292 to 5182 (From Wang and He, 1998. © 1998 lEEE).

The number of dipoles in the dipole layer is one important factor affecting the inverse solution for cortical-potential imaging. Fig. 4 shows the relative errors between the estimated cortical potentials and the actual cortical potentials when different number of dipoles were used. The dipole layer is a spherical surface at the radius of 0.75. The dipoles are uniformly distributed over this surface for all cases. A total of 128 electrodes, uniformly located over the upper hemisphere of the scalp, were used in evaluating the scalp potentials. The truncation level in the SVD algorithm was taken in such a way that makes the relative error

101

the smallest for different sources as well as different noise levels. Two radial dipoles, located at (±r· sin(1l'/6),0,r 'cos(1l' /6)) with varying eccentricity r, were used as the source. The number of dipoles used here are 310, 1292, and 5182. Fig. 4 shows the result when 10% Gaussian White Noise (GWN) was added to the scalp potential to simulate noisecontaminated measurements. The results show that when the dipole number is increased from 310 to 1292, the relative error decreases significantly. However, when 5182 dipoles were applied, no obvious improvement was achieved. This suggests that 1292, the number of dipoles, is an appropriate parameter for dipole layer at the radius of 0.75 [Wang & He, 1998].

Spatial resolution is the main concern in cortical-potential imaging. The spatial resolution is defined as the minimum distance between two radial dipole sources, which can be distinguished from the estimated cortical-potential images. The separation of the two areas of activity is determined by detecting whether there is a valley between the two peaks in the cortical potential distribution. The spatial resolution which the present procedure can achieve in imaging and localizing cortical sources is illustrated in Fig. 5. The horizontal axis refers to the eccentricity of the dipole sources, and the vertical axis to the spatial resolution when the radius of the head is taken as 1.0. Different curves correspond to different dipole layers. Fig. 5 shows the resolution when the noise level is 10%. In each case, an optimal truncation level was used to get the estimated cortical potentials; and 128 electrodes were employed. Fig. 5 demonstrates that, when the noise level is 10%, the present procedure can achieve a resolution in localizing cortical sources of 1.29 cm, provided the radius of the head is taken as 9.2 cm. The estimation of spatial resolution of EEG based on the CIT method is comparable to previous estimations based on scalp fields by other investigators.

Fig. 6 shows an example of superficial source imaging. The dipole layer was at a radius of 0.75 with 1292 dipoles and 128 electrodes were used. In Fig. 6, three maps - (a) the scalp potential map with 5% noise, (b) the cortical potential distribution, and (c) the reconstructed cortical potential distribution, are depicted. Fig. 6 is the result for four radial

0.3

" i ! 0.2

--0-- Dipole Layer 0.75 _ Dlpole •• yer 0.86 - Dipole Layer 0.865

0.1 +-------------,----------~ 0.7 0.8 0.9

Source Eccentricity

Figure 5. Spatial resolution of the spherical CIT procedure versus eccentricity of actual sources when the noise level in the scalp recorded potential is 10 percent. Different dipole layers were used. For the dipole layer at the radius of 0.75, 1292 dipoles were used. For the dipole layer at the radius of 0.85, 5182 dipoles were used. For the dipole layer at the radius of 0.865, 5182 dipoles were used (From Wang and He, 1998. © 1998 IEEE).

102

[a] [b] [c]

Figure 6. An example of cortical potential imaging for cortical radial sources. The eccentricity of the dipole sources is 0.75. The dipole layer was at the radius of 0.75, and 128 electrodes were used in the inverse procedure. (a) shows the scalp potential map with noise level of 5%, (b) shows the actual cortical potential distribution, and (c) shows the reconstructed cortical potentials. Notice that the four extrema corresponding to the four radial dipoles are well reconstructed. The potential distributions over the upper hemisphere were projected top down to the x-y plane by setting z = O. The map corresponds to a squared area, which contains the projection ofthe upper hemisphere (From Wang and He, 1998. © 1998 IEEE).

dipole sources at locations (±0.75·sin(Jl"/7),0,0 .75·cos(Jl"/7» and (O,± 0.75· sin(Jl" /7),0.75 ·cos(Jl" /7». Notice that even with noise, the present CIT procedure can get a reasonably good estimation of the poles of the cortical potential distribution, and provide spatial information about the underlying sources, which are indistinguishable in the scalp potential maps.

4.2.3 CIT in a Realistically Shaped Inhomogeneons Head Model

Head geometry is another important factor, which plays an important role in high resolution imaging of brain electric activity. Extension of the CIT to a realistically shaped inhomogeneous head model has been made by several investigators independently.

Our CIT method based on a spherical inhomogeneous head model would provide reasonable results for some clinical applications [Lian et aI, 1999a, 1999b]. However, incorporation of head geometry into the cortical imaging methods is needed to achieve the objective of imaging cortical potentials over the realistic brain surface. After our development of the 3-spheres CIT technique, we further extended our method to a 3-shell realistically shaped head model by means of BEM. The 3-shell head model is illustrated in Fig. 7. The dipole layer consists of 252 dipoles radially oriented with respect to the dipole layer, which is 10 mm underneath the cortical surface. In this work, the transfer matrix connecting the dipole moments to the scalp potentials is constructed by solving the forward problem of the lead field by a dipole located inside a 3-shell realistically shaped head model [Hamalainen and Sarvas, 1989].

Fig. 8 shows an example of the realistic geometry CIT as applied to visually evoked potentials measured in our laboratory at the University of Illinois at Chicago in a healthy human subject [Wu et aI, unpublished data]. The study was conducted according to a protocol approved by the Institutional Review Board of the University of Illinois at Chicago. Compared to the scalp potential distribution (Fig. 8(a», the reconstructed cortical potential distribution (Fig. 8(b» shows an enhanced pattern of activity over the occipital cortex.

103

~'-----I Figure 7. Schematic illustration of the realistic geometry CIT model. The brain electrical sources

are represented by a dipole layer underneath the cortical surface, located in a 3-shell realistically shaped inhomogeneous head model.

(a) (b)

Figure 8. An example of scalp potential field (a) and the estimated cortical potential field (b) using the realistic geometry CIT technique. The fields were induced by visual flash stimulation in a healthy human subject. Note the localized area of negative activity on the occipital cortex in the cortical potential map.

104

The work by Babiloni et al., which was published in 1997, is described below. The head volume conductor is represented by a 3-compartment realistic geometry model. The realistic geometry of the head model was constructed from magnetic resonance (MR) images of the subject's head. The source model was formed by a layer of 364 equivalent current dipoles with unitary moment that were radially oriented with respect to the triangle panels of the dura matter compartment. The dipole layer was placed 4-10 mm beneath the dura matter compartment. The transfer matrix from the dipole strengths to the scalp potentials were evaluated by means of the BEM [Hamalainen and Sarvas, 1989], and the TSVD algorithm was used to solve equation (4.2). The cortical potentials were further calculated from the reconstructed dipole moments.

Babiloni et al tested their realistic-geometry CIT method on simulated potentials generated from radially, obliquely, and tangentially oriented equivalent dipoles. Comparison between the forward simulated potentials, which were computed on the realistically shaped MR-constructed head model of one subject, with the inverse cortical potentials, which were reconstructed using the CIT algorithm, had suggested the feasibility of the method. Their method was also applied to experimental recordings on the scalp induced by median nerve electrical stimulation. The cortical potential estimates were calculated on the dura matter compartment for the N20-P20, P22, N24, and N30, and were integrated with the MRconstructed subject's cortical surface. The N20-P20 showed reversed frontal-positive and parietal-negative maxima contra-lateral to the median nerve stimulation, while the P22 was represented by a circumsribed area of maximum positivity in the region of the contra-lateral central sulcus approximately corresponding to the hand area. Compared with the N20-P20, the N24 extended on the contra-lateral frontal region. The N30 presented two frontalnegative and parietal-positive maxima on the contra-lateral frontal-lateral area, and a negative maximum on the frontal-mesial area.

Zanow (1997) and co-workers developed another realistic geometry CIT method, which does not assume the orientation of the dipoles over the dipole layer, embedded inside a 3-shell realistically shaped head model. Computer simulations were carried out using dipole sources at varying depths and comparisons were made among the CIT results, equivalent source imaging results, and the surface Laplacian results.' Experimental tests were also made in SEP data, and promising results were reported.

4.3. Cortical Potential Imaging via Transfer Function

4.3.1 Introduction

For the goal of cortical imaging to reconstruct cortical potentials, it would be desirable to link the cortical potential directly to the scalp potential, without an intermediate equivalent source layer such as the CIT methods described above. Similar to the CIT algorithms, a volume conductor model is needed to establish the relationship between the cortical potentials and the scalp potentials. The challenge is to relate the potential fields over two surfaces (cortical and scalp surfaces) in a volume conductor, instead of relating the scalp potential field to a dipole (or point current) source inside the volume conductor. Two types of methods have been developed to establish this relationship: finite element method [Le & Gevins, 1993; Gevins et al, 1994] and boundary element method [Srebro et aI, 1993; He et al, 1997a, b, 1998b], which will be described in detail below.

105

4.3.2 Cortical-Potential Imaging by Means of Finite Element Method

Gevins and co-workers [Le and Gevins, 1993; Gevins et aI, 1994] presented an elegant spatial enhancement procedure, called Deblurring, which used a realistic biophysical model of the passive conducting properties of each subject's head to estimate a potential distribution at the cortical surface. A finite element model of each subject's scalp, skull and cortical surface was constructed from their magnetic resonance images, and it was partitioned into three parts Sscalp, ScorticalUScutoff, and the remaining elements. Sscalp represents the elements over the scalp, and ScorticalUScutoff represents the elements over the cortical surface and the horizontal cut-off surface connecting the scalp and the cortical surface. The finite element method is applied to Poisson's equation with the following boundary conditions:

oVU(x,y,z) en == 0 onSSCalp

u(x,y,z) = G(x,y,z) on ScorticalUScutoff

(4.3)

(4.4)

where 0' is the conductivity, u(x, y, z) is the potential distribution function, n is the normal direction to the scalp surface, and G(x, y, z) is the potential distribution function at the cortical and cut-off surfaces. The following matrix vector relation is then obtained

Cu=o (4.5)

where C is the global conductance matrix and u is the vector representation of the continuous potential distribution function u(x, y, z) defined in the Deblurring region. When u is partitioned into the aforementioned three parts, denoted by u/, Uz, and U3, respectively, and the matrix C is partitioned correspondingly, equation (4.5) becomes

(4.6)

which can be rearranged and reduced into

(4.7)

Here A is the transfer matrix relating scalp potential u/ to cortical potential U3. For an arbitrarily complex Deblurring region, the matrix A may not be invertible. That requires the Deblurring method be implemented as a multidimensional optimization scheme that searches for an optimal cortical potential distribution whose forward solution best fits the measured scalp potential distribution. In their study, the search starts with Go(x, y, z), computes the corresponding forward solution uo(x, y, z), and measures the goodness-of-fit between uo(x, y, z) and the measured U(x, y, z) at the scalp surface. GJ(x, y, z) is then sought by the optimization scheme using the goodness-of-fit measure and the process is repeated. The iterative process is stopped at step i when G;{x, y, z) produces a forward solution that is acceptably close to U(x, y, z).

Gevins et al validated their Deblurring method in an epileptic patient undergoing direct cortical recordings (1994). The 14.92 Hz evoked potential was elicited by stimulation of right-hand fmgers, and recorded from the scalp. The cortical potentials were then

106

estimated by using their deblurring method, and compared to the cortical potentials directly recorded in the same patient. The original scalp EP distribution shows a single, near circular area of maximal potential in the mid-parietal area of the left hemisphere. The deblurred cortical EP shows a more compact and irregularly shaped maximum in about the same area, with a polarity reversal towards the midline. The recorded actual cortical EP shows a single, even more compact maxima in about the same area. The computed cortical EP better matches the actual cortical EP than does the original raw EP recorded at the scalp.

In Gevins and co-workers' study, the inverse reconstruction of the scalp potential distributions across realistic head models has been performed with finite element methods. Compared to the boundary-element technique, their method presented an important advantage and some disadvantages. The advantage is that it can take into account the local anisotropy within each compartment of the head model. As the result, the modeling of local head conductivity is much more efficient with the finite element methods, if such detailed conductivity is known for a human subject. Considering that values for tissue resistivities including their anisotropies are only approximately known [Nunez, 1987b], the potential advantage of the FEM-based cortical imaging may not be realized until there is a breakthrough on the noninvasive estimation of the tissue resistivity and anisotropy. Also the solution of this system needs to be calculated on powerful computers that can handle a large mass of data. Furthermore, the inverse reconstruction of the cortical potentials with the finite element method uses non-linear mathematical optimization procedures that imply a continuous rearrangement of the mathematical modeling of the data when different potential distributions are processed.

4.3.3 Cortical-Potential Imaging by Means of Boundary Element Method

An alternative approach to the cortical-potential imaging in a realistically shaped head model can be achieved by means of the boundary element method (BEM). The application of BEM in bioelectric inverse problems has decades of history. As early as the 1970s, Barr and co-workers (1977) pioneered the inverse reconstruction of epicardial potentials from the body surface potentials in a realistically shaped heart-torso model, by means of the BEM. In the field of cortical-potential imaging, two works have been reported in applying the BEM to reconstruct the cortical potentials from the scalp potentials via the transfer function approach: (1) BEM-based cortical-potential imaging in a homogenous head model [8rebro et al, 1993], and (2) BEM-based cortical-potential imaging in an inhomogeneous head model [He et al, 1997a,b,1998b].

4.3.3.1 Cortical-Potential Imaging in a Homogeneous Head Model

8rebro et al. (1993) applied the BEM algorithm to estimate the cortical potentials from the scalp potentials in a homogeneous head model of realistic shape. Two closed surfaces were selected within a homogeneous volume conductor of arbitrary geometry such that no current sources exist between tl)em. The two surfaces represent the scalp (8) and the brain (B) surfaces. Application of Green's second identity to the homogeneous volume conductor enclosed by the scalp and cortical surface yields the following

Ps.~rlIs +~~BrlIB + GsnrB =0

PBSrllS +PBBrlIB +GBBrn =0 (4.8)

107

where Cl> sand Cl> H are vectors of potentials on the scalp and brain surfaces respectively, r H is

the normal component of the gradient of Cl> H , and P and G are the matrices depending on the geometry and the passive electrical properties of the head. A similar algorithm developed by Barr et al (1977) was used to construct the transfer matrixes. If the surfaces are smooth, the matrices Pss, PBB and GBB are strongly diagonal and can be approximated by the identity matrix. Equation (4.8) can be written in the form

(4.9)

where U=[I-GSBPBSrl[GSB-PSB].

Equation (4.9) can be solved for Cl>H using Lagrangian multipliers and a constrained least squares minimization technique [Srebro et al, 1993] incorporating the constraint that the solution be as smooth as possible. Lagranginan minimization yields

(4.10)

The operator matrix R asserts the smoothness constraint on the solution. If j represents the node and k represents an adjacent node, R(j, k) is the negative reciprocal of the distance between j and k, and R(j, j) is the sum of the reciprocal distances between the node and all its adjacent nodes. y is a constant, which can be found by visual examination, and it is chosen just large enough to minimize obvious artifacts.

The above method was tested using a tank model of the human head, so that the calculated and observed "epicortical" potential fields could be compared under conditions that simulate a living head. Experimental YEP tests were also performed in human subjects. Obtained results show that the observed epicortical fields are more restricted than the scalp fields, spreading about 3 times less. However, the lack of significant low-conductivity inhomogeneity introduces considerable numerical errors in this method, as with any other homogeneous head model.

4.3.3.2 Cortical-Potential Imaging in an Inhomogeneous Head Model

We have developed a new BEM based cortical-potential imaging method in our laboratory at the University of Illinois at Chicago, in which both the realistic geometry and the inhomogeneity of the head can be taken into account [He et al, 1997 a,b, 1998b].

We modeled the head as a multi-compartment model, or piece-wise homogeneous model, in which each region is assumed to be isotropic and homogeneous. As the 3-shells model is widely accepted in the field of the brain inverse problem, we implemented our approach [He et aI, 1997a, 1998b] in a 3-shell realistically shaped inhomogeneous head model (Fig. 9), with the three shells representing the scalp, the skull and the brain tissue, respectively. Assume each shell is homogeneous and different shell has different conductivity. Since brain electrical sources exist only in the brain, there are no active sources in the volume VI and V2. So Green's second identity can be applied to VI and V2, separately. Applying to the volume VI results in

108

~PI +~2U2 +G12r 2 =0

P21 U I +P22U2 + Gn r2 =0 (4.11)

whereUk is the column vector consisting of potentials at every surface element on 8k, and r k

is the column vector consisting of au / (Jr" at every triangle element on 8k but just inside of VI. P and G are coefficient matrices.

Ii

SI O~

S3

Figure 9. A 3-shell volume conductor model. The volume between SI and S2 is denoted as V .. and that between S2 and S3 is denoted as V2•

Similarly, applying Green's second identity to the volume V2 between 82 and 8] can produce

p;p; +P2PJ -G22r; +G2Jr J =0

P'P; + P,PJ + GJ2r~ + GJJrJ = 0 (4.12)

where U2 is the column vector consisting of potentials at every surface element on 82 and r; is the column vector consisting of au / (Jr. at every triangle element on 82 but just inside of V2. The boundary conditions on 82 can be expressed as follows,

(4.13)

From equations (4.11)-(4.13), the brain surface potential U] can be related to the scalp potential U/ by

(4.14)

Using equation (4.14), one can calculate the scalp potential from the cortical potential distribution. In practice, the vector of the measured scalp potentials U is a subset of the potential vector U/ in equation (4.14). Therefore, the scalp potential measurement can be connected with the brain surface potential by the submatrix of ~3' T:

U=T·UJ (4.15)

In our approach, we use the general inverse method (Section 3.2) to reconstruct the cortical potential from the scalp potential. As a result, the brain surface potential can be obtained by

109

(a) (b)

Figure 10. An example of the cortical potential inverse solution (b) corresponding to four radial dipole sources. The scalp potential map is shown in (a), to which 10% GWN is added. Note the cortical potential map can reconstruct well the extrema of the 4 radial dipole sources, whereas the scalp potential map fails.

(4.16)

where U3 is the estimated cortical potential field. We have tested our algorithm by using a three-shell concentric sphere head model

because of the availability of analytic solution. Fig. 10 shows an example of the inverse image reconstruction using our BEM based algorithm (He et ai, 1997b). Figs. 10(a)-(b) are the results for four radial dipoles, where (a) is the scalp potential distribution (with 10% GWN), and (b) is the reconstructed cortical potential distribution using the general inverse estimation. Fig. 10 demonstrates the feasibility of reconstructing cortical potentials from simulated noise-contaminated scalp potentials.

(a) (b)

Figure 11. An example of cortical potential distribution (a) estimated from SEPs over the scalp (b), about 20 ms after stimulation of the right median nerve in a human subject. The cortical potentials were reconstructed based on a 3-shell realistic geometry head model using the boundary element technique, which we have recently developed in our laboratory. Notice the dramatically improved spatial distribution of the potential over the cortex as compared to the potential over the scalp (From He, 1998. © 1998 IEEE).

110

Fig. 11 shows an example of the cortical potential image (a) estimated from the scalp SEP (b) in a human subject [He, 1998]. The scalp SEP data were provided by F Babiloni, which were collected using 111 electrodes 20 msec following somatosensory stimulation at the right median nerve in a human subject. The truncated SVD algorithm was used to perfonn the inverse imaging. Fig. 11 clearly indicates that the cortical-potential image provides a much more focused pattern of activity as compared to the scalp potential map. Activities are observed in the cortical potential image over the somatosensory cortex as expected [Towle et al, 1995]. On the other hand, the scalp potential map shows a widely diffused pattern of activity over the scalp, not revealing the spatial detail.

4.4 Equivalent Surface-Source Imaging

Imaging of brain electrical activity has been perfonned in the source space since early explorations on brain electrical imaging [Freeman, 1980; Hamalainen & Ilmoniemi, 1984; Nunez, 1987a]. In such approaches, the specific source distribution is usually assumed at first, and the solution of Poisson's equation in a head volume conductor model is detennined which can best account for the measured scalp data. Source models include a planar current dipole layer [Freeman, 1980; Hamalainen & Ilmoniemi, 1984], current dipole sources underneath scalp electrodes [Nunez, 1987a], and a T-shape dipole layer simulating ~ortical neural current sources [Wang et al, 1992]. Recently, a fine dipole layer over the cortex, with orientation of the dipoles being the nonnal to the cortex at each position, has been used to reconstruct cortical sources from noninvasive recordings [Srebro, 1997; Philips et al, 1997]. The shape of the cortex was detennined from MR images of the subject, and the inverse solution leads to the estimate of the cortical sources corresponding to the scalp potentials.

We have fonnulated an equivalent surface-source imaging approach to add to the family of brain electrical imaging [He et al, 1998a]. We attempt to eqUivalently represent the 3-dimensional brain electrical activity by means of a 2-dimensional surface source inside the brain. Fig. 12 illustrates schematically our equivalent surface-source model. In a volume conductor model representing the head, a double or single layer is considered, which is assumed to enclose all principal brain electrical sources. There are two points where our fonnulation differs with other reported approaches: 1) we explicitly consider the source layer (double or single layer) instead of the potential on a surface; 2) the surface source model represents equivalently the electrical activity inside the whole brain, instead of local areas occupied by the source layer. The closed equivalent dipole layer in the CIT methods [He et al, 1996; Zanow, 1997; Wang & He, 1998] may be considered as a realization of the equivalent surface-source model, though no attempt is being made to construct the cortical potentials from the equivalent sources in the proposed equivalent surface-source imaging approach.

As discussed in Section 4.2.2 on the CIT algorithm in a 3-spheres head model, we are able to place the equivalent dipole layer at a position close to the epicortical surface. Such capability is important since the equivalent surface-source distribution should be able to enclose in principle all brain electrical sources in order to realize equivalent source modeling using a double-layer or a single-layer [Stratton, 1941]. We have studied the application of directly using the distribution of the dipole layer strength, instead of the cortical potential, to perfonn equivalent-source imaging [He et al, 1998a]. Using the closed dipole (double) layer, we have found that the analytical solution of the strength of the dipole (double) layer is proportional to the electrical potential over the same surface when the exterior space outside of the dipole layer is replaced by air [Yao & He, 1999]. Our finding provides the theoretical basis for evaluating the equivalent dipole (double) layer imaging results.

111

Figure 12.

S scalp

Primary Source (3D)

Seq

Equivalent Source (2D)

A schematic diagram illustrating the concept of the equivalent surface-source model. In this model, the 30 primary current sources are equivalently represented by a 20 double-layer or a 20 single-layer, which encloses the 30 primary sources inside the brain.

Fig. 13 shows an example of the equivalent dipole layer imaging results [He, 1998]. Fig. 13 (a) shows the distribution of the analytic solution of the dipole layer strength. Fig. 13 (b) shows the inversely estimated distribution of the dipole layer strength at an eccentricity of ' 0.80 which clearly reveals existence of 4 radial sources from noise-contaminated scalp potential data shown in Fig. 13(a). The zero-order Tikhonov regularization using the CRESO algorithm was used to obtain Fig. 13(b). Fig. 13 indicates the feasibility of equivalent dipole layer imaging of multiple sources in the brain.

(a) (b)

Figure 13. An example of equivalent dipole-layer imaging of cortical sources. (a) The "true" distribution of the strength of dipole-layer at eccentricity of 0.8; (b) The distribution of the strength of the dipole layer reconstructed from scalp potentials using the CRESO algorithm. (From He, 1998. © 1998 IEEE)

In the equivalent surface-source imaging approach we proposed, the equivalent surface source can also be a single layer surface source model. The single layer source model can be considered as a layer of point current source over the surface. Such a single layer source model has previously been applied in bioelectricity to equivalently represent bioelectric sources [He et ai, 1995; Yao, 1996]. By constructing a closed single-layer source

112

inside the brain, the strength of the equivalent single-layer is found to be proportional to the current density when the exterior space outside of the single layer is replaced by a perfect conductor [Yao & He, 1999]. The distribution of the single layer strength can be estimated from the scalp potentials and used to image brain electrical activity.

Fig. 14 shows examples of the theoretical distribution of (a) the equivalent doublelayer and (c) the equivalent single-layer, corresponding to two current dipoles; and (b) the equivalent double-layer and (d) the equivalent single-layer, corresponding to four point current sources. The dipole parameters are the same as those in Fig. 2. Comparing with Fig. 2 indicates that the equivalent surface source distributions provide enhanced imaging capability as compared to the scalp potential map.

The equivalent surface-source imaging approach has the advantage of less computation, yet still achieving the similar goal of cortical-potential imaging. The equivalent surface-source imaging approach can be extended to a realistically shaped head volume conductor model, in which the solution of the strength of the equivalent surface-source will need to be evaluated numerically in a realistic geometry homogeneous volume conductor model with the exterior space outside of the closed surface-source being replaced by air or perfect conductor.

( .. ~., ......... ; ,~"

" •• '. ~: '4. (c" •

(a) (b)

(c) (d)

Figure 14. Equivalent double-layer strength (a) and equivalent single-layer strength (c) representing two radial dipoles located at (0.6cos(n/6), ±0.6sin(n/6), 0); Equivalent double-layer strength (b) and equivalent single-layer strength (d) representing two point current sources located at (0.7cos(n/6), ±0.7sin(n/6), 0) and two point current sinks located at (0.scos(n/6), ±0.Ssin(n/6), 0). The radius of the equivalent double or single layer is 0.8. The equivalent layer strength is obtained by setting the conductivity outside the double layer to 0, and the conductivity outside the single layer to infinite.

113

Acknowledgement

We would like to thank Yunhua Wang, with whom some of the work reviewed in this chapter were conducted, Jie Lian for assistance in producing Fig. 13, and Fabio Babiloni, who provided us the l11-ch SEP data to test our cortical imaging techniques.

This work was supported in part by NSF CAREER Award BES 9875344, a grant from the Whitaker Foundation, a grant from the Pittsburgh Supercomputing Center through the NIH National Center for Research Resources grant 2 P41 RR06009, to Bin He.

REFERENCES

Babiloni, F., C. Babiloni, F. Carducci, L. Fattorini, P. Onorati, and A. Urbano, 1996, Spline Laplacian estimate of EEG potentials over a realistic magnetic resonance-constructed scalp surface model, Electroencephalography and clinical Neurophysiology, 1-11.

Babiloni, F., C. Babiloni, F. Carducci, L. Fattorini, C. Anello, P. Onorati, and A. Urbano, 1997, High resolution EEG: a new model-dependent spatial deblurring method using a realistically-shaped MR-constructed subject's head model, Electroenceph. & clin. Neurophysi., 102: 69-80.

Baillet, S. and L. Gamero, 1997, A Bayesian approach to introducing anatomo-functional priors in the EEGIMEG inverse problem, IEEE Trans. On Biomed. Eng., 44: 374-385.

Barr, R.C., M. Ramsey, III, and M.S. Spach, 1977, Relating epicardial to body surface potential distributions by means of transfer coefficients based on geometry measurements, IEEE Trans on Biomed Eng, 24: 1-11.

Colli-Franzone, P., L. Guerri, B. Taccardi, and C. Viganotti, 1985, Finite element approximation of regularized solutions of the inverse potential problem of electrocardiography and applications to experimental data, Calcolo, XXII: 91-186.

Crowley, C.W., Greenblatt, R.E., and I. Khalil, 1989, "Minimum norm estimation of current distributions in realistic geometries," In: Advances in Biomagnetism, S.l Williamson, M. Hoke, G. Stroink, and M. Kotani (Eds), Plenum, New York, pp. 603-606.

Cuftin, B.N., 1995, A method for localizing EEG sources in realistic head models, IEEE Trans. Biomed. Eng., 42: 68-71.

Eshel, Y., 1993, Correlations between anatomical asymmetries in the head and scalp potential amplitude asymmetry - a mathematical model, Ph.D. Dissertation, Tel-aviv University .

Fletcher, IF., A. Amir, and D.L. Jewett, 1995, Improved method for computation of potentials in a realistic head shape model, IEEE Trans. Biomed. Eng., 42: 1094-1104.

Freeman, W., 1980, Use of spatial deconvolution to compensate for distortion of EEG by volume conductor, IEEE Trans. on Biomedical Engineering, 27: 421-429.

Gencer, N.G., Y.Z. Ider, and S.J. Williamson, 1996, Electrical impedance tomography: induced-current imaging achieved with a multiple coil system, IEEE Trans. Biomed. Eng., 43: 139-149.

Gencer, N.G. and SJ. Williamson, 1998, Differential characterization of neural sources with the bimodal truncated SVD pseudo-inverse for EEG and MEG measurements, IEEE Trans. Biomed.Eng., 45: 827-837.

Geselowitz, D.B., 1960, Multiple representation for an equivalent cardiac generator, Proc. IRE, 48: 75-79.

114

Geselowitz, D.B., 1967, Bioelectric potentials in an inhomogeneous volume conductor, Biophys. J., 7(1): 1-11.

Gevins, A., 1. Le, N.K. Martin, P. Brickett, 1. Desmond, and B. Reutter, 1994, High resolution EEG: 124-channel recording, spatial deblurring fUld MRI integration methods, Electroencephalography and clinical Neurophysiology, 90: 337-358.

Golub, G.H. and C.F. Van Loan, 1989, Matrix Computations, 2nd Edition, Johns Hopkins Uni. Press, Baltimore.

Gorodnitsky, I.F., J.S. George, and B.D. Rao, 1995, Neuromagnetic source imaging with FOCUS: a recursive weighted minimum norm algorithm, Electroenceph. & clin. Neurophysiol., 95: 231-251.

Grave de Peralta Menendez, R., O. Hauk, S. Gozalez Andino, H. Vogt, and C. Michel, 1997, Human Brain Mapping, 5: 454-467.

Gulrajani, R.M., 1998, Bioelectricity and Biomagnetism, Wiley, New York. Hamalainen, M. and R. Ilmoniemi, 1984, Interpreting measured magnetic fields of the brain:

estimates of current distributions, Helsinki University of Technology Report TKK-FA559.

Hamalainen, M. and R. Ilmoniemi, 1994, Interpreting magnetic fields of the brain: Minimun norm estimates, Med Bioi. Eng. Comput., 32: 35-42.

Hansen, P.C., 1990, Truncated singular value decomposition solutions to discrete ill-posed problems with ill-determined numerical rank, SIAM J. Sci. Stat. Com put. , 11: 503-518.

Hansen, P.C., 1992, Analysis of discrete ill-posed problems by means of the L-curve," SIAM Rev., 34: 561-580.

Hansen, P .C. and D.P. O'leary, 1993, The use of the L-curve in the regularization of discrete ill-posed problems, SIAM J. Sci. Comput., 14(6): 1487-1503.

He, B., T. Musha, Y. Okamoto, S. Homma, Y. Nakajima, and T. Sato, 1987, Electrical dipole tracing in the brain by means of the boundary element method and its accuracy, IEEE Trans. Biomed. Eng., 34: 406-414.

He, B. and T. Musha, 1992, Equivalent dipole estimation of spontaneous EEG alpha activity: two-moving dipole approach, Medical & Biological Engineering & Computing, 30: 324-332.

He, B. and R.1. Cohen, 1992, Body surface Laplacian ECG mapping, IEEE Transactions on Biomedical Engineering, 39: 1179-1191.

He, B., 1994, On the Laplacian inverse electrocardiography, Proceedings of the Annual International Conference of IEEE Engineering in Medicine and Biology Society, 145-146.

He, B., Y.B. Chemyak, and R.I. Cohen, 1995, An equivalent body surface charge model representative three-dimensional bioelectrical activity, IEEE Trans. On Biomed Eng., 42(7): 637-646.

He, B., Y. Wang, S. Pak, and Y. Ling, 1996, Cortical source imaging from scalp electroencephalograms, Med. & BioI. Eng. & Comput., 34/SuppI, part 2: 257-258.

He, B., Y. Wang, and D. Wu, 1997a, Imaging Brain Electrical Activity Using a 3D Realistically Shaped Inhomogeneous Head Model, Proceedings of the Annual International Conforence of IEEE Engineering in Medicine and Biology SOCiety, 1167-1169.

He, B., Y. Wang, and D. Wu, 1997b, Cortical electrical imaging from scalp EEG using a realistically shaped inhomogeneous head model, Medical & Biological Engineering & Computing, 35/Suppl Part 2: 670.

115

He, B. and D. Wu, 1997, A bioelectric inverse imaging technique based on surface Laplacians, IEEE Trans. On Biomed Eng., 44: 529-538.

He, B., 1998, High resolution imaging of brain electrical sources, IEEE Engineering in Medicine and Biology, 17(5): 123-129,1998.

He, B., D. Yao, 1. Lian, and D. Wu, 1998a, High resolution EEG: a new surface-source imaging technique, Submitted for pUblication.

He, B., Y. Wang, and D. Wu, 1999, Estimating Cortical Potentials from Scalp EEG's in a Realistically Shaped Inhomogeneous Head Model, IEEE Trans. On Biomed. Eng., 46: 1264-1268.

He, B., in press, Brain Electric Source Imaging: Scalp Laplacian Mapping and Cortical Imaging, Critical Reviews in Biomedical Engineering.

Helmholtz, H. von, 1853, Uber einige Gesetzeder Verbeitung elektrischer Strome in Koperlichen Leitem mit Anwendung auf die theorischelektrischen Versuche, Ann. Physik. u Chem., 89: 211-233, 353-377.

Hjorth, B., 1975, An on-line transformation of EEG scalp potentials into orthogonal source derivations, Electroenceph. Clin. Neurophysiol., 39: 526-530.

Hobson, E.W., 1931, The theory of spherical and ellipSOidal harmonics, Cambridge University Press, Cambridge.

Homma, S., T. Musha, Y. Nakajima, Y. Okamoto, S. Blom, R. Flink, K.E. Hagbarth, and U. Mostrom, 1994, Location of electric current sources in the human brain estimated by the dipole tracing method of the scalp-skull-brain (SSB) head model, Electroencephalography and clinical Neurophysiology, 91: 374-382.

Jackson, J.D., 1975, Classical Electrodynamics, Second Edition, John Wiley & Sons. Jeffs, B., R. Leahy, and M. Singh, 1987, An evaluation of methods for neuromagnetic

imaging reconstruction, IEEE Trans. On Biomed. Eng., 34: 713-723. Kavanagh, R.N., T.M. Darcey, D. Lehmann, and D.H. Fender, 1978, Evaluation of methods

for three-dimensional localization of electrical sources in the human brain, IEEE Trans. Biomed Eng., 25: 421-429.

Khosla, D., M. Singh, and M. Don, 1997, Spatio-temporal EEG source localization using simulated annealing, IEEE Trans. Biomed. Eng., 44: 1075-1091.

Kirsch, A, 1996, An Introduction to the mathematical theory of Inverse Problems, SpringerVerlag, New York.

Lawson, C.L. and RJ. Hanson, 1974, Solving Least Squares Problems, Prentice-Hall, Englewood Cliffs, N1.

Le, J., V. Menon, and A Gevins, 1992, Local estimate of surface Laplacian derivation on a realistically shaped scalp surface and its performance on noisy data, Electroenceph. Clin. Neurophysiol., 92: 433-441.

Le, 1. and A Gevins, 1993, Method to reduce blur distortion from EEG's using a realistic head model, IEEE Trans. on Biomedical Engineering, 40: 294-299.

Lian, J., D. Yao, and B. He, 1998, A New Method for Implementation of Regularization in Cortical Potential Imaging, Proceedings of the Annual International Conference of IEEE Engineering in Medicine and Biology SOciety, 2155-2158.

Lian, J., F. Babiloni, J. Hughes, J. Fino, D. Wu, and B. He, 1999a, Cortical Imaging of Human SEP Activity and Epileptic Foci Using a 3-Spheres Inhomogeneous Head Model, In: Proc. Of the 3rd International Workshop on Biosignal Interpretation, B. He, K. Yana, M. Akay, S. Cerutti (Eds.), pp. 155-157.

116

Lian, 1., K. Spencer, 1. Dien, E. Donchin, and B. He, 1999b, Exploration of P300 and novelty P3 components by means of the cortical imaging technique, Medical & Biological Engineering & Computing, 37/Suppl. 2: 1634-1635.

Malmivuo, J. and R. Plonsey, 1995, Bioelectromagnetism, Oxford University Press. Miller, K., 1970, Least squares methods for ill-posed problems with a prescribed bound,

SIAM J. Math. Anal., 1: 52-74. Morozov, V.A., 1967, Choice of parameter for the solution of functional equations by the

regularization method, Sov. Math. Doklady, 8: 1000-1003. Morozov, V.A., 1984, Methods for Solving Incorrectly Posed Problems, Springer-Verlag,

Berlin. Mosher, 1.C., P.S. Lewis, and R.M. Leahy, 1992, Multiple dipole modeling and localization

from spatio-temporal MEG data, IEEE Trans Biomed Eng., 39: 541-557. Nunez, P., 1981, Electric Field of the Brain, Oxford University Press, London. Nunez, P., 1987a, Removal of reference electrode and volume conductor effects by spatial

deconvolution of evoked potentials using a three concentric sphere model of the head, The London Symposium, Electroenceph. clin. Neurophysiol., 39/[Suppl]: 143-148.

Nunez, P., 1987b, A method to estimate local skull resistance in living subjects, IEEE Trans. Biomed Eng., 34: 902-904.

Nunez, P., R.B. Silibertein, P.J. Cdush, R.S. Wijesinghe, A.F. Westdrop, and R. Srinivasan, 1994, A theoretical and experimental study of high resolution EEG based on surface Laplacian and cortical imaging, Electroenceph. and Clin. Neurophysiol., 90: 40-57.

Oster, H.S. and Y. Rudy, 1992, The use of temporal information in the regularization of the inverse problem of Electrocardiography, IEEE Trans. Biomed Eng., 39(1):69-75.

Pascual-Marqui, R.D., C.M. Michel, D. Lehmann, 1994, Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain, Int J Psychophysiol., 18: 49-65.

Peralta-Menendez, R.G., and S. L. Gonzalez-Andino, 1998, A critical analysis of linear inverse solutions to the neuroelectromagnetic inverse problem, IEEE Trans. On Biomed Eng., 45: 440-448.

Perrin, F., O. Bertrand, and J. Pemier, 1987, Scalp current density mapping: value and estimation from potential data, IEEE Trans. Biomed Eng., 34: 283-288.

Philips, 1.W., R.M. Leahy, J.C. Mosher, and B. Timsari, 1997, Imaging Neural Electrical Activity from MEG and EEG, IEEE Trans. Med Imaging, 16: 338-349.

Plonsey, R. and D.B. Heppner, 1967, Considerations of quasi-stationarity m electrophysiological systems, Bull. Math. Biophys., 29: 657-664.

Plonsey, R., 1969, Bioelectric Phenomena, McGraw-Hill, New York. Plonsey, R. and R.C. Barr, 1986, Bioelectricity - A Quantitative Approach, New York

Plenum Press. Press, W.H., S.A. Teukosky, W.T. Vetterling, B.P. Flannery, 1992, Numerical Recipes in C,

Cambridge University Press. Rush, S. and D.A. Driscoll, 1969, EEG electrode sensitivity - An application of reciprocity,

IEEE Trans. on Biomedical Engineering, 16: 15-22. Scherg, M. and D. Von Cramon, 1985, Two bilateral sources of the AEP as identified by a

spatio-temporal dipole model, Electroencephal. and Clin. Neurophysiol., 62: 32-44. Schwan, H.P. and C.F. Kay, 1956, Specific resistance of body tissues, Circ. Res. 4(6): 664-

70. Schwan, H.P. and C.F. Kay, 1957, Capacitive properties of body tissues, Circ. Res. 5(4): 439-

43.

117

Shim, Y.S. and Z.H. Cho, 1981, SVD pseudo-inversion image reconstruction, IEEE Trans. Acoust, Speech, Signal Processing, 29: 904-909.

Sidman, R.D., V. Giambalvo, T. Allison, and P. Bergey, 1978, A method for localization of sources of human cerebral potentials evoked by sensory stimuli, Sensory Processes, 2: 116-129.

Sidman, R, M. Ford, G. Ramsey and C. Schlichting, 1990, Age-related features of the resting and P300 auditory evoked responses using the dipole localization method and cortical imaging technique, J. Neuroscience Methods, 33: 23-32.

Sidman, R., D. Vincent, D. Smith, and L. Lee, 1992, Experimental tests of the cortical imaging technique--applications to the response to median nerve stimulation and the localization of epileptiform discharges, IEEE Trans. on Biomedical Engineering, 39: 437-444.

Smyth, W.R, 1968, Static and Dynamic Electricity, 3rd Edition, McGraw-Hill, New York. Song, A.C.K. and Z.J. Koles, 1995, Principal components localization of the sources of the

background EEG, IEEE Trans. Biomed Eng., 42: 59-67. Srebro, R, Oguz, R.M., Hughlett, K., and P.D. Purdy, 1993, Estimating regional brain

activity from evoked potential field on the scalp, IEEE Trans. Biomed. Eng., 40: 509-516.

Srebro, R and R.M. Oguz, 1997, Estimating cortical activity from VEPS with the shrinking ellipsoid inverse, Electroenceph. & din. Neurophysi., 102: 343-355.

Stratton, J.A., 1941, Electromagnetic Theory, McGraw-Hill, New York. Sullivan, B.J. and B. Liu, 1984, On the use of singular value decomposition and decimation

in discrete -time band-limited signal extrapolation, IEEE Trans. Acoust. Speech, Signal Processing, 32(6).

Tikhonov, A.N. and V. Y. Arsenin, 1977, Solutions of Ill-posed problems, Wiley, New York. Towle, V.L., S. Cohen, N. Alperin, K. Hoffmann, P. Cogen, J. Milton, R. Grzeszczuk, C.

Pelizzari, I. Syed, and J.P. Spire, 1995, Displaying electrocorticographic findings on gyral anatomy, Electroencephalography and clinical Neurophysiology, 94: 221-228.

Twomey, S., 1963, On the numerical solution of Fredholm integral equations of the first kind by the inversion of the linear system produced by qudarature, J. Assoc. Com put. Mach., 10: 97-101.

Uutela, K., M. Hamalainen, and R. Salmelin, 1998, Global optimization in the localization of neuromagneitc sources, IEEE Trans Biomed Eng., 45: 716-723.

Van Veen, B.D., W.V. Drongelen, M. Yuchtmn, and A. Suzuki, 1997, Localization of brain electrical activity via linearly constrained minimum variance spatial filtering, IEEE Trans Biomed. Eng., 44: 867-880.

Wang, J.Z., S.J. Williamson, and 1. Kaufman, 1992, Magnetic source images determined by a lead-field analysis: the unique minimum-norm least-squares estimation, IEEE Trans. Biomed Eng., 39: 665-675.

Wang, Y. and B. He, 1998, A computer simulation study of cortical imaging from scalp potentials, IEEE Trans. Biomed. Eng., 45: 724-735.

Wu, D., J.P. Saul, and B. He, 1995, Epicardial Inverse Solutions from Body Surface Laplacian Maps: A Model Study, Proceedings of the Annual International Conference of IEEE Engineering in Medicine and Biology Society, 227-228.

Wu, D., Y. Wang, and B. He, unpublished data. Yamashita, Y., 1982, Theoretical studies on the inverse problem in electrocardiography and

the uniqueness of the solution, IEEE Trans. Biomed Eng., 29: 719-725.

118

Yao, D., 1996, The equivalent source technique and cortical imaging, Electroenceph. & clin. Neurophysiology, 98: 478-483.

Yao, D. and B. He, 1998, The Laplacian weighted minimum norm estimate of threedimensional equivalent charge distribution in the brain, Proc. of the Annual Int. Con! of IEEE Engineering in Medicine and Biology Society, 2108-2111.

Yao, D. and B. He, 1999, Forward theory of equivalent surface-source imaging, Submitted for publication.

Zanow, F., 1997, Realistically shaped models of the head and their applications to EEG and MEG, Ph.D. Thesis, University of Twente.

Zhao, F., DN. Wu, G. Krumick, D. Wu, and B. He, 1998, A New Realistic Geometry Spline Laplacian Algorithm and its Application to YEP, Proc. of the Annual Int. Conf. of IEEE Engineering in Medicine and Biology SOciety, 2183-2185.

119

APPLICATIONS AND CONTROL OF HIGH VOLTAGE PULSE DELIVERY FOR

TUMOR THERAPY AND GENE THERAPY in vivo

Berti) R.R. Persson

Department of medical radiation physics Lund University Hospital S-221 85 Lund, Sweden

HISTORICAL INTRODUCTION

External electric fields applied to biological materials interact directly on free electric charges such as electrons and ions as well as on ionic groups in larger molecules. They also interact with dipoles such as water and induce dipoles in molecules with polarizable groups. The cell membranes seem to be the critical target for interaction causing intra-molecular transitions and intermolecular processes that lead to structural reorganization of the cell membranes. Applying high voltage pulses to cells in cultures or in tissue cause various degrees of structural reorganizations in the cell membranes, which might end with a dielectric collapse or breakdown. This condition is either called electroporation or electropermeabilization and can be reversible or irreversible depending on the characteristics of applied voltage pulses (Zimmermann and Neil, 1996; Zimmermann, 1982; Zimmermann et al. 1981; Zimmermann et al. 1974; Stampfli, 1958). In such a transient state, the membrane become permeable to molecules that normally doesn't pass this barrier into the cytoplasm of the cell. This condition can be used for direct transfer of genes, other nucleic acids, proteins, and other molecules into cells and microorganisms. Another possibility is that neighbor cells with membranes in transient state might fuse together and form a new giant cell. These properties of membranes in transient state has led to electric field pulse techniques which have gained increasing importance in cellular and molecular biology, in gene technology and in various medical therapeutic procedures (Neumann et al. 1989). The effect of permeabilizing cell membranes by applied electric pulses is widely used in biochemistry, genetics and cell biology to introduce exogenous, membrane insoluble molecules into cells. The requirement· for high efficiency of electropermeabilization and molecular transfection depends ont4e membrane dielectric properties, cell-shape and size, as well as pulse parameters such as shape, amplitude, length and number of pulses.

Advances in Electromagnetic Fields in Living Systems, Vol. 3 Edited by J .C. Lin, K1uwer AcademicIPlenurn Publishers, 2000 121

THEORETICAL ASPECTS

A cell membrane is usually modeled as a bilayer lipid membrane that is an effective barrier for passage of ions and hydrophilic molecules. Such a membrane has an electrical specific conductance Gm to ions such as Na+ and K+ is typically 10-8 S.cm-2 or smaller (Tien, 1974). For cell membranes the electrical conductivity is generally higher, in the order of 10-3

S cm·2 (Cole, 1972). When a high-voltage pulse is applied the membrane specific conductance increase dramatically and reach values as high as 1 S cm-2 (Kinosita and Tsong, 1979; Hibino et al. 1991). The resting electrical potential for a cell membrane is normally around -10 mV and the threshold for dielectric breakdown is in the order of 150-500 mY. Assuming a thickness of the bilayer of 5 nm the electric field strength over the membrane is in the order of 30-100 106 Vim (300-1,000 kV/cm). Cell membranes can sustain as much as 1 000 m V of applied transmembrane potential II membr. i. e. an electric field strength of 2000 kV/cm, when microsecond to millisecond high-voltage pulses are used (Coster and Zimmermann, 1975; Sale and Hamilton, 1968). The dielectric strength of cell membranes depends both on the amplitude and on the length of the applied electric field (Neumann et at 1989).

Equivalent electrical circuit model

In Figure 1 is shown the experimental set up for high voltage impulse treatment of tumors or other tissues. Electrodes are configured on both side of the tumor. They can be arranged in various patterns in order to cover the tumor with a homogeneous electric field. The electrical impulses are generated in a power supply, with a control unit. A measuring unit is required for controlling the electrical parameter of the tissue that are feed back into the control unit for performing an optimal high voltage pulse treatment of the tumor.

High Voltage Power

K~==:)Control Unit ~==:)

lectrod s

Tumor

Measuring Unit

Figure 1. High Voltage power supply, Control unit, and measuring unit for performing of electric pulse treatment of a tumor in its surrounding tissue.

The diagram of Figure 2 shows the equivalent circuit of the parallel-conductance model of the electroporated membrane. The high voltage pulses are generated by a current source I(t) in parallel with the resistor RN . The cell membranes are represented by the capacitor CM in parallel with the resistors lIgM and I/G(VM,t). The conductance G(VM,t) is a voltage dependent conductance that reflects the dielectric breakdown of the membrane electrical conductance of the cell membrane. The normal leakage conductance of the

122

membrane is represented by ~. When the switch is closed, current flow through the extracellular space and the voltage, VM is generated over each cell, and they would affect the conductivity of cell membrane. The resistor RE represents the resistance of the extracellular space, electrodes and wires.

By applying the voltage VM the current through the membrane iM increase linearly with time, that reflects the dielectric breakdown of the membrane. The rate of current increase depends in a logarithmic and square manner of the trans-membrane voltage VM according to the equation (Chemomordik and Chizmadzhev, 1989)

(1)

where A" and Bo are constants In exponential form this equation can be written as

diM V, (p V,2) -== ·a·exn . dt M r M (2)

where the constant a == Ao . exp(- Bo)

U(t) .. • VM .. 1IG(V"t) 1IG(V"t)

ic 119M

R,

C iE

c

RE

Figure 2. Equivalents circuit where high voltage pulses is generated by U(t). The extracellular space is represented by the resistor RE • The cell membranes are represented by the capacitor C in parallel with the resistor l/G(t), where G(t) is the electrical conductivity of the cell membrane due to electropermeabilization and &.! is the normal passive conductivity. The intracellular space is represented by the resistor RIo

From the parallel-conductance equivalent circuit, (Figure 2) the trans-membrane current iM can be determined from the equation:

123

(3)

Applying a square pulse one can assume that VM is constant during the pulse. Thus the time derivative of the membrane current iM is

(4)

Thus, the rate of dielectric breakdown is given by the equation:

dG =a.ex/n.V?) dt \P M

(5)

where the constants were derived from the results of Chemomordik et al. 1989 by (Krassowska, 1995; Chemomordik and Chizmadzhev, 1989).

a=2.5·10-3 [mS .. cm-2.ms-1 ] ;P=2.5.10-5 [my2]

The trans-membrane voltage VM in an applied electric field E is given by the equation

(6)

where Aip M is the membrane potential in absence of external electric fields and is assumed independent of E and of time.

Applying a rectangular pulse, the buildup of interfacial polarization is given by:

The stationary value of /:lip can be derived from the cell membrane modeled as a thin spherical shell of thickness d and outer radius a, and the definition of the electric field applied. (Figure 3)

The stationary value of /:lip thus obtained is given by the equation

/:lip = -3/2· E .a·lcosol· j(;') (8)

where f{A.) is a conductivity term. For most cases of interest the radius of the cell a is much larger than the thickness of

the membrane. The conductivity of the cytoplasm ;'0 is equal to the conductivity of the medium ~ that is much larger than the conductivity of the membrane ~. Then

j(;') = [1- exp(- 2Ant I3RC,.)] (9)

where Cm is the membrane capacitance.

124

Since the membrane capacitance em does not change much, the kinetics of the increase of the potential of the membrane mainly depend on the conductivity of the medium Ito.

Thus wij:h j(1t);:::1 the stationary value of VM in the direction of the applied electric field vector E is given by the equation:

(10)

Note that coso /icosol = + 1 for the right hemisphere in Figure 3 and cos 0 /lcOS 01 = - 1 for the left hemisphere.

Since the energy of a deformed charge distribution is proportional to the square of the electric field, the deformation energy W is roughly proportional to E2 particularly if E·a»llrpM·

If we assume a first order kinetics and N is the number of unaffected cells in a sample the number of cells dN undergoing electric breakdown in the time interval (I, I+dl) is

+ + +

+ +

-dN=k·N·dl

Applied Electric Fieled

E r

+, , , + " -, , , , ,, ,

r' , , , , , , , , , , aJ,L -T!:;r

'---~----t'1:~~I

(11)

Figure 3. Interfacial polarization of a spherical non-conductive shell of thickness d and outer radius Fa in a constant external field E. The lower diagram shows the change of potential !J.rp over the membrane.

The rate constant k depends on the activation-energy Wa according to the Ahrenius equation k = A . exp[B . Wa / T) where A and B are constants and T is the absolute temperature oK.

Thus, the rate of electric breakdown per cell can be written as follow:

125

dG = a. exp(p. V~)"., a'· exp(b'. E 2 )"., a"· exp(b". Wa / r)= k dt (12)

Within a limited temperature interval that is considered as constant in most experimental and in vivo conditions the electric breakdown depend in an exponential way on the square of E i.e. the electric energy.

The following expression is obtained after integrating during pulses of certain duration , applied on a sample with No cells before the treatment and N unaffected cells after the treatment (Tomov, 1995).

In(No/ N) = am + b"'. E2 r (13)

The duration r after treatment with n exponential pulses is r = RC . n and with n square formed pulses oflength t is r = t . n .

Molecular and cellular effects

The molecular basis for transiently membrane breakdown by pulsed electric fields is still unclear, and the chemical nature of the chemical species involved are not fully clarified (Zimmermann, 1982). Recently, it has been shown that electroporation of cells induce luminescence which indicate that membrane peroxides are formed in cells after electroporation (Maccarrone et al. 1995b). Erythroleukemia K562 cells subjected to poreforming electric fields showed a parallel increase in membrane hydroperoxide content and permeability values (Maccarrone et al. 1995a). Hydroperoxide induce remarkable structural changes in synthetic lipid layers. Thus the formation of hydroperoxide might be one of the chemical processes leading to breakdown of the membranes of animal and plant cells treated with high voltage pulses (Wratten et al. 1992). Other recent observations suggest that lipid peroxidation might be the basis for electropermeabilization of plant protoplasts and Chinese hamster ovary cells (Gabriel and Teissie, 1994; Biedinger et al. 1990).

EFFECT OF HIGH VOLTAGE PULSES ON CELL SURVIVAL

The effect of applying different number of high voltage pulses of various shape, amplitude and duration has been studied in vitro on lung fibroblast V79 cells, and in vivo on tumors implanted on the flank of rats.

In vitro treatment of Fibroblast cells

The cell killing effect achieved by applying high voltage pulses of various shape, amplitude, duration and number on lung fibroblast V79 cells in vitro have been studied by (Danfelter et al. 1998). Lung fibroblast cells from V79 Chinese hamsters with a doubling time of 9 to 12 hours were used. The cells were grown as a monolayer in DulbeccO's modified eagle medium (Life Technologies). The cellsuspension was incubated at 37°C in humidified air with 5% CO2 for 5 days in a Nunclon flask. A cell suspension with 106 cells per ml was used for the experiments. The cells were held at room temperature (20°C) for at least 30 min before the experiment begun.

In Figure 4 is given the survival function, of V79 cells treated with exponentially decaying high voltage pulses. The regression equation of 93 experiments (R = 0,79; p<O.OOOI).

126

(14)

Regression forced through zero of the same data gives the equation:

(15)

In Figure 5 is given the survival function, of V79 cells treated with square formed high voltage pulses. The regression equation of 43 experiments (R = 0,81; p<O.OOOI)

SOO

400

";' 300 1/1

";'

C; ~~ 200

Z ""'0

z :s- 100

0

In[No/ N] = -43(± 31)+ 3.4 .1O-4 (± 4 .10.6 ). E2 t·n

o 200000 400000 600000 800000 1000000 1200000 1400000 1600000

E2 / V2.cm·2

(16)

Figure 4. Survival of V79 cells (1 000 000 cells/ml) treated with exponential high voltage pulses of time constant T(s), number n, and fieldstrength E (V /cm). Linear regression of all 93 experiments. The dotted lines are 95% confidence limits.

Regression forced through zero of the same data gives the equation:

(17)

The parameters are in good agreement with no significant difference between exponential and square formed pulses. Thus, the theoretically derived expression seems to fit quite well with experimental findings of in vitro cell culture.

127

In vivo effect of High Voltage pulses only

The effect of high voltage pulses on in vivo growth of tumors has been investigated by exposing N32 brain tumors implanted under the skin of the thigh of Fischer-344 rats (Engstrom et al. 1998). Tumors were exposed to high voltage pulses of exponential decay of 1300-1400 V/cm with a time constant of 1 ms. The tumors were gently fixed in position between two external flat electrodes that were mounted on a slide caliper and connected to a pulse generator. Sixteen pulses were delivered transdermally to the whole tumor during 20 seconds by two rectangular steel electrodes of 2x2 cm mounted on a slide caliper. The tumors were subjected to four sessions of 16 pulses each on four consecutive days, each session lasting for approximately 20 seconds. The tumor size was estimated by measuring the width and length of the tumor with a slide caliper and calculating the area as an ellipse. In Figure 6 the estimated area of the tumors is displayed for controls and the average for the tumor treated with high voltage pulses in two different experiments.

1000

900

800

700 '7

." 600

'7 500 c:: '71-: 400 Z -'0 300 Z

:5 200

100

0

o 500000 1000000 1500000 2000000 2500000

Figure 5. Survival of V79 cells (1 000 000 cells Iml) treated with square form high voltage pulses of various length T (s), number n, and fieldstrength E (V I em). Linear regression of all 43 experiments. The dotted lines are 95% confidence limits.

Almost immediately following the pulse treatment, edema was developed that lasted for approximately 2-3 days. Then 2-3 days after the last treatment session extensive tumor necrosis followed. The flexibility of the treated leg was reduced for a short period after the treatment but was later fully recovered. The study includes two experiments using essentially identical treatment protocols.

In the first experiment, 2 out of 5 treated animals showed complete remission five days after the end of the treatment. After 100 days, however, there was only one rat alive with complete remission. The average time the tumor reached the critical size of 3 x3 cm2 was 21 ± 5 days that is almost twice as long as the average time of 13±3 days for the 8 controls.

A two-tailed t-test assuming equal variance is significant with p=0.02 but with unequal variance p=O.07 it is hardly significant.

In the second experiment with 8 controls and 4 treated rats, one rat treated with high voltage pulses was sacrificed after 100 days and no regrowth of tumor was found. The

128

group of 4 treated animals reached the critical tumor size of 3 x3 cm2 after about 60 days that is almost three times as the average time of 20B days for the 8 controls.

A two-tailed t-test assuming equal variance is significant with p=0.002 but with unequal variance p=O. 07 it is hardly significant. As a conclusion there is an effect on tumors treated with high voltage pulses only, that gives a 2-3 times delay in the tumor growth and gives durable therapeutic effect in 2 out of 9 treated animals (22%).

1 ....

1D

N um E E - !ro IV

I!! c:( @ .. 0 E .... .=

200

~ after inocubiim. / days

Figure 6. Estimated area of the tumors for controls and the average for the tumor treated with high voltage pulses in two different experiments.

HIGH VOLTAGE IMPULSE TREATMENT COMBINED WITH CYTOTOXIC DRUGS

The method of permeabilizing cell membranes with high voltage pulses has also a potential for use in tumor therapy by enhancing the effect of cytotoxic drugs (Poddevin et al. 1991; Mir et al. 1988; Okino and Mohri, 1987). Numerous subsequent studies have confirmed the increase in cytotoxicity in electroporated cells exposed to anticancer drugs both in vitro and in vivo (Gehl et al. 1998; laroszeski et al. 1997; laroszeski et al. 1997; laroszeski et al. 1996; Sersa et al. 1997; Sersa et al. 1997; Sersa et al. 1995; Tounekti et al. 1993; Belehradek et al. 1993; Belehradek et al. 1991; Dev, 1994; Dev and Hofmann, 1994; Mir and Orlowski 1999). Clinical trials in cancer patients have been performed in order to increase the intracellular level of chemotoxic agents in tumors by using electroporation in vivo (Heller et al. 1996; Heller, 1995; Heller et al. 1995; Glass et al. 1997).

Animal studies

The first attempt to apply high voltage pulses to brain tumors was done by Salford et al. 1993 who managed to prolong the survival of RG2 glioma bearing Fischer-344 rats with 200% by iv. administration of bleomycin followed by intra-cranial treatment with exponential decaying high voltage pulses (Figure 7).

129

IN VIVO ELECTROPORATION 1993 Bleomycin + Electroporation ( 800V/cm )

Per -cent surv ivals

10

&poReIIIIai pUIu. u....,.,., ... at o. t "'"

Treated 23 animals

Intracerebral RG2tumours

..... _-- 2 animals: 30 + days ~

20 30 Time after treatment I days

10-13 days a"er Inoculalion or 5000 cells

40

Figure 7. Survival of RG2 glioma bearing Fischer-344 rats: 18 control animals and 23 animals after iv. administration of bleomycin followed by intra-cranial treatment with exponential decaying high voltage pulses (Salford et al. 1993).

Jaroszeski et aL (1997a) treated rat hepatomas with high-voltage pulses and reported 84.6% objective response rate compared to 10% in the control groups that received partial or no treatment.

Nanda et al. (1998) treated human pancreatic tumors (Pan-4-JCK) implanted subcutaneously onto nude mice with high-voltage pulses and various cytotoxic agents (Bleomycin, Carboplatin, Mitomycin C). The drug was slowly injected at the base of the tumor and after a time laps of 5-1 0 min the animals was treated with high-voltage pulses. At day 89 after treatment, 6 out of 8 tumors (75%) treated with high-voltage pulses combined with Bleomycin showed complete response. At the same time only 2 out of 8 tumors (25%) treated with high-voltage pulses combined with mitomycin C, and only 1 of 8 treated (13%) with high-voltage pulses combined with carboplatin showed complete response (Nand a et aL 1998a).

Human larynx tumors, Hep-2 implanted in nude mice were treated with high-voltage pulses after intratumoral injection of Bleomycin (Nanda et al. 1998b). On day 67 after the treatment, 10 of 12 treated animals were in complete remission, while all mice in the control groups had progressive disease. These results might indicate that High-voltage impulse therapy has potential for treating laryngeal tumors.

Significantly prolonged survival was observed in BALB/c mice with subcutaneous colorectal carcinoma (CRC) tumors after treatment with intratumoral injection of bleomycin and electroporation with 8 pulses with a field strength of 1000 V/cm and 99 Ils pulse length. Survival rate varied between 6-12 days in the controls and between 10-24 days in the treated animals (Kuriyama et al. 1999).

Wi star rats with a colon adenocarcinoma transplanted into the liver were treated with intratumoral injection of bleomycin (0.3 IU, equivalent to ca 160 Ilg) combined with electropermeabilization. The high voltage pulse treatment applied 16 exponentially decaying electric pulses of 1300 V/cm and a time-constant of 1 ms. This treatment resulted in significantly reduced tumor volume (p<O.OOI) and 92 % cure rate (12 out of 13, p<O.OOI) compared to the sham groups. Measurements of macro phages and T -lymphocytes (CD4 and CD8) in and around the tumors indicate that the electro chemotherapy treatment stimulate the hosts immune system (Engstrom et al1999d,e).

130

Clinical trials

Head and Neck tumors. The first clinical trial of using high-voltage impulses was conducted by Mir et al. (1991) on patients who had already undergone conventional cancer treatment for head and neck squamous cell carcinomas. The treatment protocol involved intravenous injection of Bleomycin (10 mg/m2) followed by pulsing the tumor with caliper electrodes. Of the 40 nodules treated, 23 nodules were in complete regression, 6 nodules were in partial regression, and 11 nodules without response.

Panje et al. (1998) used high-voltage pulses in combination with intratumoral injection of 1-5 units of Bleomycin per cm3 of tumor in 10 patients with head and neck cancer (8 squamous cell carcinomas, 1 adenoid cystic carcinoma, and 1 adenocarcinoma).

The results of this phase IIII trial show that intralesional bleomycin injection combined with high-voltage pulse treatment produces head and neck cancer necrosis. The overall clinical response rate> 50%, tumor reduction, in this study was 80%. Forty percent of the patients had biopsy-proven complete response at a mean follow up of 10 month.

They concluded that high-voltage impulse therapy in combination with intralesional Bleomycin offers a new method for treating head and neck cancer. Simplicity, tumor specificity, and minimal side effects are encouraging factors for continued development of this method to a clinical routine.

Dermal tumors. Mir et al. (1991) treated 28 squamous cell cancer dermal metastases with high-voltage pulses and bleomycin and found 72% local tumor necrosis and 50% complete response rate. In another study Belehradek et al. (1993) treated 8 patients with a total of 40 dermal squamous cell cancer metastases utilizing intravenous bolus injection of Bleomycin (10 mg/m2) followed by High-voltage pulse treatment to the tumor nodules. Twenty-three out of the 40 nodules (57%) demonstrated a complete response (Belehradek et al. 1993).

Glass et al. (1996,1997) demonstrated a response-rate greater than 90% in treatment of skin cancers, including basal cell carcinoma and cutaneous melanoma with intralesional Bleomycin and high-voltage pulses (Heller et al. 1996; Glass et al. 1997; Glass et al. 1996; Glass et al. 1996).

Rudolf et al. 1995 reports on clinical experience from treatment of malignant melanoma patients with High-voltage pulses and bleomycin (Rudolf et al. 1995).

HIGH VOLTAGE IMPULSE TREATMENT COMBINED WITH RADIATION AND RADIONUCLIDE THERAPY

High voltage pulses combined with external and brachy radiation therapy

Although a lot of progress is made in cancer research, radiation therapy is still the most widely used therapy modality for cancer treatment. Many attempts are made to develop this technique and to enhance the sensitivity of radiation resistant tumor. The survival of tumor cells treated with high voltage pulses seems to mimic the survival of radiation treated cells. Thus, one feels tempted to study if treatment with high voltage pulses has a sensitizing effect on radiation therapy (Engstrom et al. 1999c,d).

In the experiments, male rats of the Fischer-344 strain were used with rat glioma N32 tumors implanted subcutaneously on the flank or on the thigh of the hind leg. The tumor is only weakly immunogenic and grows readily intracerebrally, subcutaneously and in vitro (Siesjo et al. 1993). Tumors were produced by injecting 100 000 N32 tumor cells just below the skin. Tumors were treated about 4 weeks after inoculation when a solid tumor has developed with a diameter of 1-1.5 cm. Before pulse treatment of the tumors, animals

131

were anesthetized with 5% chloral hydrate given intraperitoneally (i.p.). The fur over the tumor was shaven and carefully covered with electrocardial paste to ensure good electrical contact between electrodes and skin. The paste also moistened the skin and prevented skin necrosis from the pulse treatment.

Animals were arranged into five groups. One group of 8 controls (C) received no treatment. Single treatment with radiation was given to 5 rats (XR. T) and treatment high voltage pulses only to 4 rats (HVT). The combined treatment was done in two groups, 4 rats were given high voltage pulses followed by radiation (HVT+XR.T) and 5 rats radiation treatment followed by high voltage pulses (XR.T+HVT). The total absorbed radiation dose given to the tumors was 20 Gy, given in 4 fractions of 5 Gy each day in four consecutive days. Radiation treatment (XR.T) was conducted 10 to 15 minutes before the treatments with high voltage pulses (HVT). The time interval between the end of the pulsation of the tumor and radiation treatment was 4 to 5 minutes.

In Figure 8 is shown the effect of single type of treatment on the growth of N32 tumors implanted on the flank of rats. Controls (C) received no treatment. Two groups received single treatment with either radiation treatment only (5 Gy in four consecutive days), or high voltage pulses only EP=1300V, 16 pulses in 4 consecutive days.

WI , , , ,

lim , , , .,

d E 1m [] Controls ~= 6.6 days E , • Radiation ~= 11 .5 days ,

41 <;) ,

E a EP ~= 11.1 days , , :l , '0 V > ..., ~

0 E ~ 1m

o

o 2 4 6 8 ro u w ~ m ~ ~ u ~ ~ ~ ~ ~ Thne afta"' sta1: .. treatnut / days

Figure 8. Effect of single type of treatment on the growth of N32 tumors implanted on the flank of rats. Controls (C) received no treatment. Two groups received single treatment with either radiation treatment only (5 Gy in 4 consecutive days), or high voltage pulses only EP=1300V, 16 pulses in 4 consecutive days. .

In Figure 9 is shown the effect on the tumor size of combined treatment in 4 consecutive days first with 6°Co-radiation 5 Gy and then 16 high voltage pulses (1300 Vlcm, 1 ms) (XEP). The effect of combined treatment in 4 consecutive days first with 16 high voltage pulses (1300 Vlcm, 1 ms) and then with 60Co-gamma radiation 5 Gy (EPX) is shown as well.

Tumor response was evaluated by measuring the length, width and thickness of the tumor and estimating the tumor volume as an ellipsoid. Animals (single-treated and controls) were sacrificed when the tumor had reached a size of 5 cm3. That occurred after: 42±27(SD) days for the 12 animals treated with high voltage pulses, and 17±5(SD) days for the 16 untreated controls p<0.02 (t-test uneq.var). Corresponding analysis for 9 paired samples, 42±27 days for treated rats and 15±4 days for 9 untreated paired controls (p=O.OI

132

paired samples). All treated animals showed an initial partial or complete tumor remission within a few days after the end of the 4-day treatment. Six out of the 9 treated animals had complete remission with no sign of tumor recurrence after 100 days.

ME E QJ

E ::l "0 > ... o

1000

800

600

400

E 200

~

o

I

I

• EPX ,.

'V XEP

• Controls I

I

I • , ,

o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34

Time after start of treatment I days

Figure 9. Effect of combined treatment on the growth of N32 tumors implanted on the flank of rats. Controls (C) received no treatment. Effect on the tumor size of combined treatment in 4 consecutive days first with 16 high voltage pulses (1300 V / cm, 1 ms) and then with 60 Co-gamma radiation 5 Gy (EPX).

Results from combined treatment with 60Co_garnma radiation therapy (RT) and high voltage pulse therapy (HVT) show that the average time when the tumor had reached a size of5 cm :

> 94 days; when treated first with R T then HVT > 1 00 days; when treated first with HVT then R T

All treated animals showed an initial partial or complete tumor remission within a few days after the end of the 4-day treatment. For the combination radiation 4x5Gy and then 4x HVT (1300 V/cm), 3 out ofthe 11 treated animals had complete remission> 100 days.

For the combination 4xHVT (1300 V/cm) and then radiation 4x5Gy, 3 out of the 4 treated animals had complete remission> 1 00 days.

The outcome of this preliminary study of combining high voltage pulses and radiation therapy suggests an astonishingly effective therapeutic response (Engstrom et al 1999c,d). It is thus worthwhile to further explore this combined type of treatment to improve the therapy of resistant tumors. It might be of particularly interest to apply high voltage impulses combined with brachy therapy where the radioactive source (i.e. iridium) can serve as electrode.

High Voltage Impulse Therapy Combined with Radiophamaceuticals and RadionucIide therapy.

Gamma-camera imaging and evaluation of 99Tcm_DTPA uptake. It has been shown that the accumulation of 99Tcm -DTP A in rat muscle tissue treated with high voltage pulses can be evaluated by the use of a gamma camera (Engstrom et al. 1999a,d). The diagnostic radiopharmaceutical 99Tcm -DTP A . (Diethylenetriamine-pentaacetic acid) can is

133

used as a model radionuclide to study how various parameters affect the electropermeabilization efficiency of the treatment. It can also be used as a tracer mixed with the therapeutic compound (cytostatic drugs or Auger electron emitting drugs).

Thus in combination of High Voltage pulses with radionuclide therapy one gets enhanced therapeutic effect both by increased cellular uptake and by radio sensitization of the target tissue.

A gamma camera (General Electric Maxicamera II, GE Company, USA) together with a graphical software (Nuclear MAC, Scientific Imaging, Littleton CO, USA) were used to view and evaluate the 99Tcffi_DTPA uptake in the various region of interest ("ROI") of the animals, with and without electropermeabilization treatment. The gamma camera is equipped with a scintillation NaI(TI) crystal viewed by 37 photomultiplicator tubes which record the gamma quanta of 140 keY energy emitted in the nuclear decay of 99Tcrn_DTPA in the rats. The spatial resolution is about 5 mm at the gamma energy of 140 keY. Gamma camera measurements of the activity uptake in the treated regions should be performed 5 to 8 hours after treatment. By this time the animals has excreted most of the activity in the circulation and there are no remaining effects due to the muscle contraction induced by the high voltage pulses. In Figure 10 is shown a gamma camera picture of the 99Tcm_DTPA uptake in the region treated with high voltage pulses.

Effect of Electric Field strength. The uptake of a 99Tcrn_DTPA in muscle tissue of rats was studied following in vivo treatment with high voltage pulses. The treatment parameters that have been studied are pulse shape, field strength, pulse duration, number of pulses and time interval between drug administration and electric pulse delivery. The activity of 99Tcrn -DTP A uptake in electropermeabilised tissue was measured at various field strength ranging from zero to 1600 V/cm.

Figure 10. Gamma-camera picture of 99Tcm _DTPA uptake in the region treated with high voltage pulses ("HVP"). The kidneys and bladder are filled with 99Tcm _DTPA cleared from the blood during the examination.

At constant field strength, the uptake of 99Tcffi-DTPA increases rapidly on pulse duration up to 1-2 milliseconds for both exponential and square pulses as shown in Figure

134

11. Eight pulses were given with field strength of 600 V/cm with either exponential pulses or square wave pulses. There is no significant difference in uptake between the two pulse forms when pulse length or time-constant are identical.

The energy delivered by the electrical pulses is given by the following equation:

Where n is the number of pulses i, E; is the field strength during the pulse i (Ti is the electric conductivity of tissue and Pi is the tissue density

11

10

9 • Data points

r-- Linear Ftt: Y = (42+ -3)E-4*E , G/ 8 , ,

.\I: - - - - . Upper 95% Confidence Limit ca - - - - Lower 95% Confidence Limit , .. 7 , C. , ::J , ,

~ 6

'> 5 , , :;::; , U --ca ~

4

~ 3 --&! 2 -

--0

,

0 200 400 600 800 1000 1200 1400 1600

Applied Field Strength I V.em -1

,

1800

(18)

Figure 11. The dependence of relative drug uptake with field strength up to 1600 V /cm for square wave pulses. The results displayed indicate following linear dependence: Relative uptake = (42±3)·10·4.E; 01 /cm)

The results displayed in Figure 11 indicate a linear dependence in drug uptake with field strength up to 1600 V/cm for square wave pulses. A significant drug uptake recorded above 400 V/cm but no significant electropermeabilization is observed for field strengths below 200 V/cm.

Electropermeabilization with both square wave pulses and exponential pulses was performed to compare the permeabilization efficiency and to determine which pulse-shape yields the highest drug uptake after in vivo electropermeabilization.

The ratio between the energy absorbed in tissue in case of exponential and square wave pulses of maximum fieldstrengths Uo,exp, and UO,.q and time"-constant 1exp and pulselength t.q, is given by the following equation. If the conductivity is assumed the same in the two cases, we can derive the following expression:

"'fU2 • -21/ t".dt W O.exp e U 2 t .....!!:E.. = ""0 ______ = O.erp· up

If"U2 A U;.sP ·2· tSq O.sq • wi

o (19)

The equation states that, the energy delivered from I200-V/cm exponential pulses of

135

1.0 ms (time constant) is the same as that from 1200-Vlcm square pulses of 0.5 ms duration. The uptake of 99Tcm_DTPA after delivery of 8 exponential pulses of 1 ms and 1200 V/cm was, however, only 1.5 times higher than for the same number of 99 Ils square pulses of the same field strength, although the delivered energy was 5 times higher. In another experiment with eight exponential pulses of 1 ms, 600 V/cm, the relative uptake was about 3.5-4,which corresponds to that obtained with 8 square pulses of 99 J.ls, 1200 V/cm. In this case, the ratio of absorbed energy Wexp1Wsq was 1.3 that is in much better agreement between electropermeabilization and absorbed energy (Engstrom et al. 1999a,d).

The conductivity that is assumed the same in deriving the energy ratio might vary differently in the two cases. These phenomena will be described in more details in section 7.

Effect of time interval between drug administration and electric pulse delivery. If the drug to be internalized is given intravenously, the cells should be perrneabilised at the time of maximum drug concentration in the interstitial fluid, which may last only for a few minutes. It is thus vital to choose the correct time delay between drug administration and electropermeabilization to obtain maximum intracellular drug delivery to the target cells. In the present study, 99Tcm_DTPA was given i.v. as a bolus at various time intervals ranging from 20 minutes before pulse delivery to 8 minutes after pulse delivery. Eight exponential pulses of 1200 Vlcm with a time constant of 1.0 ms were given to the thigh muscle with a frequency of 0.75 Hz (Engstrom et al. 1999a,d).

The variation of drug uptake with the time-interval between injection of 99Tcffi -DTPA and delivery of electric pulses is shown in Figure 12. The drug uptake in the muscle tissue after electroperrneabilization treatment indicate that maximum uptake of 99Tcffi_DTPA is obtained if electric pulse are delivered either at 2 minutes after or 2-4 minutes before i.v. injection. The diagram shows that if the time-interval between drug and pulse delivery is too short, i.e. within one minute before or after electropermeabilization of the tissue, the drug uptake is considerably decreased. Time-intervals longer than 4 minutes is less critical for the drug uptake efficiency than a too narrow time window.

In the study of the time interval between drug administration and electrical pulse delivery, we found a pronounced decrease in drug accumulation at time intervals shorter than 1 min. This might due to a vascular contraction effect by the electric pulse. Arteries and arterioles have several layers of muscle cells that contract the vessels when the vessels are exposed to stress or damage. This is called vascular spasm. Veins and venules have less layers of muscle cells (1-2) or none at all (small venules). Electric pulses of high voltage are known to stress cells and tissues. If electric pulses are given immediately before intravenous drug administration, the blood flow to the electropermeabilised site is cut off and precludes any substantial drug uptake in this area. This vascular effect can last for several minutes. On the other hand, if the electric pulses are given soon after i. v. drug delivery, the contraction of particularly the arterioles will block the blood flow resulting in a low drug concentration in the target area and low tissue accumulation (Engstrom et al. 1999a,d).

Effect of pulse length and number of applied pulse. The uptake of 99Tcffi_DTPA after treatment with 1 to 32 exponential wave pulses of 1200 V/cm at a frequency of 1 Hz indicates an initial linear correlation but beyond 8 pulses, the slope of the curve is then declining. An increase from 4 to 8 pulses leads to 50% increase in drug uptake but additional 8 pulses (totally 16) only results in an additional 25% increase in drug uptake. Thus to obtain a significantly elevated drug uptake, the tissue needs be treated with several pulses of 1200 Vlcm when using pulses of millisecond duration (Engstrom et alI999a,d).

The uptake dependence of pulse number can be compared with that of pulse duration. Cell electropermeabilization depends on increased number of pulses in a similar way as increased length of a single pulse. However, pulses longer than 4 ms to involve risk for tissue damage even at intermediate voltages (600 V/cm). Increasing the pulse number and reducing the pulse length is likely to spare the treated tissue. This is desirable especially

136

12

11

10

j 9 nr

8 -c.. => 7 >.

~ 6

n 5 « CD 4 > :;::; 3 (ij

~ 2

1

0 First HV then Injection First injection then HV

-20 -15 -10 -5 o 5 10 15

Time-interval between Injection of activity and HV-treatment I min

Figure 12. The variations of relative drug uptake with the time-interval between injection of 99Tcffi_DTPA and High-Voltage treatment with 8 exponential pulses of 1000 V fcm and time constant 1 ms.

when very large molecules such as proteins and nucleic acids are to be internalized into cells which requires electropermeabilization with longer pulses.

High voltage Impulse therapy combined with radioactive labeled cytotoxic drugs

Use of electro-pulsation in radio nuclide therapy. Intratumoural injection of radioactive compounds has gained interest in the treatment a variety of tumors. Radionudide therapy has, however, the drawback of inefficient uptake of radioactive compounds into the cytoplasm compartment of the tumor cells. By applying electropulsation with administration of a radioactive compound, the permeabilised cell membranes will result in an increased uptake and decreased leakage of the drug from the injection site to the circulation.

This phenomenon is demonstrated in Figure 13 that shows the gamma camera picture taken 3 days after i.v . irljection of lllIn-labelied bleomycin and treatment with 8 pulses of 1200 V/cm at 4 min after administration (Engstrom et aI1999d).

High voltage Impulse therapy combined with Boron Neutron Capture Therapy (BNCT)

The question whether boronated compounds accumulate sufficiently in brain tumors is of great importance with regard to Boron Neutron Capture Therapy (BNCT) for malignant brain tumors and other cancer forms. This relies on a selective uptake and accumulation of boron into the tumor cells prior to irradiation with thermal neutrons (Ceberg et al. 1992). Already in 1994, we presented our findings about the effect of high-voltage pulses in combination with i.v. administration ofborocaptate BSH in the RG2 rat glioma model. We found that the uptake in the tumor increased when high-voltage pulses were applied, but also spread in the whole brain. Using another compound boronated porphyrin (BOPP) we also found a tenfold increased in tumor uptake but no leakage to the surrounding brain (Ceberg et al. 1994).

137

Recently a Japanese study of the cell membrane permeability and the intracellular accumulation ofborocaptate (BSH) after exposure to high-voltage impulses were presented with the aim to be used in clinical BNCT in combination with Bleomycin (Ono et al. 1998). Another Japanese study of a new Boron-Porphyrin compound showed that the accumulation of boron in the tumor was too low to elicit a therapeutic effect (Shibata et al. 1998). By application of high-voltage pulses, the uptake might be increased to a therapeutically effective level.

Figure 13. Gamma camera image showing the drug retention of Ill-Indium labeled bleomycin in a N32 glioma tumor three days after treatment of electric pulses and lllIn_ bleomycin administration.

HIGH VOLTAGE IMPULSE TREATMENT FOR DNA DELIVERY IN GENE THERAPY

Delivery of DNA, either in vitro or in vivo, always requires the vectorisation of this large molecule, which normally does not penetrate the cell membrane. The high promises of gene therapy for future treatments of many diseases using the DNA as a drug has been impaired by problems linked to gene vectorisation to the target cells. Viral vectorisation present known limits (size, risks, and host immune reaction against the vector, production, and stability of the viruses) Chemical vectorisation has also identified limits (low efficiency, capture by non-target cells, and toxicity). A few "physical" methods already exist, like the ballistic approach (gene gun) which is however limited to surface cells. Therefore, there is a need for new vectorisation systems. Another way to achieve efficient DNA delivery to tissues in vivo is based on the application of high voltage pulses in vivo.

Cell electroporation has become the most widely used method for in vitro gene transfection in bacteria and in eukaryotic cells in the cell and molecular biology laboratories. At present studies are performed to study cell electropermeabilization in tissues exposed in vivo to high voltage pulses.

For gene therapy applications, survival of the cells of the tissues exposed to the electric pulses is an essential requirement. Different tissues have been treated in mice: liver, tumor,

138

muscle, using long square wave pulses of intermediate to low strength (with respect to the pulses used for electrochemotherapy) (Suzuki et al. 1998; Xie et al. 1992). Experiments demonstrate that in vivo, the EP have two roles: the target cell electropermeabilization, plus direct effects on DNA molecules (electrophoretic effects) resulting in high increase of DNA electro-transfer in the electropermeabilised cells. Therefore, the electric pulses must be much longer than those used in electro chemotherapy (and nevertheless still respect the flrst constraint mentioned here above, that is preserve cell viability). Therefore special generators delivering variable electric pulses and delivering the minimal energy necessary to achieve both permeabilization and transport must be developed. This also rises the requirement of methods to measure and control the effect on tissue of the treatment with high voltage pulses to achieve the most optimal result.

METHODS FOR OPTIMIZING AND CONTROL OF HIGH VOLTAGE IMPULSE TREATMENT

Impedance spectrometry for measurement and controlling tissue effects of High Voltage Impulse Treatment.

Living tissue is composed of cells, which are surrounded by a highly electrically insulating lipid membrane. Between the cells, there is an extracellular liquid compartment of high electric conductivity. Electrical DC and low frequency currents are limited to the extracellular space. With increasing alternating current frequency the impedance of the cell membrane, decrease. Electrical impedance readings over a frequency range then allow separation of extracellular space and of the membrane themselves.

In Figure 2 is shown an equivalent circuit of the parallel-conductance model of a cell membrane. The diagram shows the equivalent circuit where high voltage pulses are generated by U(t). The resistor ~ represents the extracellular space. The cell membranes are represented by the capacitor CM in parallel with the resistors l/G(VM,t) and l/~. The passive electric conductivity of the cell membrane is represented by ~ that mainly due to ion-channels G(VM,t) is the electrical conductivity of the cell membrane due to dielectric breakdown. The intracellular space is represented by the resistor RI . When the switch is closed current flow through the extracellular space and the EMF generated over each cell affect the conductivity of cell membrane. We will develop such models further in order to model the impedance change in tissue recorded by impedance spectroscopy.

The most widely used model describing the conduction of alternating electrical currents in tissue along intra and extra-cellular pathways is the Cole-Cole model (Cole and Cole, 1941). The model is based on measurements of the conductivity of suspensions of spherical cells and impedance measurements on frog eggs (Cole and Guttman, 1942).

This model is commonly implemented by a simple RC-circuit as shown in Figure 2 and the impedance of tissue as a function of frequency of the applied AC voltage is given by the equation.

(20)

From this model the impedance values at inflnite frequency Z", and at (j) = 0 Zo (DC) can be estimated by the equations:

. _ RE . R] + Ru . C 11m Ztissue ~ Zo - ----"'--'-----""----117-->0 R] +RM +RE (21)

(22)

139

By measuring the impedance in the low frequency domain,information will be obtained about the electrical characteristics of the tissue by extrapolating the experimental data to obtain Z", and Zo, respectively.

The relaxation time 't can also be expressed in terms of a characteristic frequency Ie = If T and the phase angle is often given as a coefficient a.

Thus, the equation fitted to experimental conditions is often given in the form:

<01

3S)

N~ 300

I 2S) i!! J2 1l

200 N Q.j 0 l3) c: (]) ..... (]) 100 :t:

i:5 3J

0 1

__ DlIlIlV

....... DlnlV __ DIIXlV

__ D«XJV

_m:xw _nov

10

Z Z Zo-Z", tissue = co + a

1+ j{~]

............ 100 11m) lwm

Frequency I Hz

(23)

Figure 14. Results from experiment with impedance measurements in the thigh of a rat after injection of 0,2 ml saline intra-muscular before and after 4 min electroporation treatment with 8 exponential pulses of 1200 V / cm with a time-constant of 1 ms.

The impedance measuring system usually consists of a signal generator with a known high output impedance, Zout. A measuring device able to measure the voltage applied by the generator, and a device used to control and the measurements, evaluate and display the measured values. The measurements of Z shown in the following diagrams are made by using a swept sine wave.

In order to simulate the conditions of real treatments 0.2-ml saline was injected intra muscularly in the thigh of a rat. After 4 min the impedance was measured before and after high voltage pulse treatment with 8 exponential pulses of 1200 V/cm and a time-constant of 1 ms. The results displayed Figure 14 clearly show a decrease in impedance over the whole frequency range from 20 Hz to 200 kHz. In this frequency range, the impedance mainly depends on the resistivity of the tissue (Engstrom et al. 1999b,d).

In Figure 15 is shown the spectral difference of impedance values "(Zbefore - Zafter)" before and after treatment with 8 exponential pulses of a time-constant of 1 ms and varying applied electric field strength (V/cm). Large fluctuations appear in the lower frequency part of the differentiated spectrum. Above a frequency of 300 Hz the value of the differential is

140

in the range of -0,05 and converges to zero at 200 kHz. Thus, the integral of the impedance difference between 300Hz and 200kHz should bea good measure for studying the electropermeabilisation effect of high voltage treatment of tissue.

2oo':!z \..l /Dz = J\Zbe!o,..(f J-ZafI.J j )pj

300Hz

700.-----------------------------------------, 800

Esoo .s::. o -!400 fi ! ..5 300

200

-.- Before treatment -e- After 8 pulses 1200 Wcm

100~~~rn~--~~~w_~~~~--~~~W_-T~rl

10 100 1000 10000 100000

Frequency I Hz

(24)

Figure 15. Difference of impedance measurements in the frequency interval 15 - 200 000 Hz in the thigh of a rat. Saline (0.2 ml) was injected intra muscular before and after 4-min electroporation treatment with 8 exponential pulses of various electric fieldstrength 011 cm) with a time-constant of 1 ms.

In Figure 16 the value of the integrated difference impedance spectrum ID. is plotted versus the applied field-strength at treatment with 8 exponential pulses of time constant 1 ms. There is a highly significant (p<001) linear relation between the value of IDz and corresponding applied field-strength:

/Dz =2.4(±0.7)+0.0176(±0.007).E (25)

/Dz = 0.02(± 0.008). E (26)

The positive offset at zero field-strength depend onthe time-delay (4 min) between the first and second measurement during which the injected bolus of saline has spread out in the tissue. For controlling purpose, the slope of 0.02 through zero should be accurate enough.

141

~ .. 35 E .s:.

0 :IE - 30

if . E 15 2 1; 8. 1AI .. II U I: ~ ~ 15 ;; II U I: 10 .. " 8. .5

5 'l;

~ II 0 :E

Integral of [Zbeto,. - Z ••• ,l between 300 Hz and 200 kHz

, , , , , ,

, "",

,

-- Linear Fit

, , , ,

, , , , ,

, "

"

----- Upper 95% Confidence Limit

- -- -- Lower 95% Confidence Limit

o 200 400 600 800 1 000 1200 1400 1600 1800 2000

Electric Field strength / V.cm· 1

Figure 16. Integrated difference Impedance spectrum lD,= Jlz __ (f)-z .... JliW lOOH.

plotted versus the applied fieldstrength (V / cm) at treatment with 8 exponential pulses of time constant 1 ms.

The results of IDz from analogue impedance measurements before and after treatment with different number of pulses and of various time-constant also indicated linear relations.

IDz = 1.7(±OA)· N; 600 V/cm, 1 ms (27)

IDz = 6.0(±0.6)· 'f 1200 V / cm, 8 pulses (28)

For in vitro conditions, we assume that the conductivity 1S constant during the treatment session and the efficiency is as follows:

(29)

For in vivo conditions, we assume the average conductivity for the treatment sequence varies inversely with applied field-strength (J = const.f E .

This will result in a linear relationship for In[No IN Invivo that is assumed proportional to IDz according to the following formula:

In[No/Nl . ex: ID = 0.002· E . T· n nVlVO Z (30)

Thus by using electric impedance over the frequency interval 300 Hz to 200 kHz it is possible to derive a quantity IDz that can be used for controlling the permeabilization effect of high voltage pulses on tissue. This concept will be further developed into more precise dosimetric methods for measuring the change of conductivity between each pulse.

142

CONCLUSION

Clinical applications of in vivo electropermeabilization that are currently used both combined with chemotherapy and as a tool for gene transfection in vivo. Development of clinical devices for application of electric pulses in vivo and use of special anticancer agents IS III progress.

The technique of in vivo electropermeabilization has several advantages in the clinic. The instrumentation required for electropermeabilization is compared to the methods presently used in tumor and gene therapy neither costly nor complicated to handle. In clinical practice, however, besides pulse generator and electrodes, it is necessary to apply dosimetry and treatment control. Impedance spectrometry measurements in the frequency range of 300 - 10 000 Hz has been shown suitable for recording the rate of occurrence of electroporation in the tissue or tumor. By using this method it is possible control the intensity of electric pulse treatment to get highest tumor cellkilling in tumor therapy, and to optimize transfection ratio and cell survival in gene therapy.

There may also be future applications for in vivo electropermeabilization in radiation therapy. In radionuclide and radioimmuno-therapy therapy, short-range p-emitting and Auger-emitting radionuclides will be internalized into the cells of solid tumors by applying electric pulses. This will increase the cellular uptake and deliver a high locally radiation absorbed dose (McLean, Blakey et aI., 1989;McLean and Wilkinson, 1989;Howell, 1992;Humm, Rao et aI., 1994). In fact, IllIn-Bleomycin, is a radio-pharmaceutical that combines radiotherapy and chemotherapy (Hou, Hoch et aI., 1985a;Hou, Hoch et aI., I 985b;Kairemo, Ramsay et aI., 1996;Kairemo, Ramsay et aI., 1997). Other such radiopharmaceuticals are 195mpt_cisplatin or 195mPt_transplatin (Howell, Kassis et aI., 1994).

Both radionuclide and radioimmuno-therapy (RIT) (i.e. the concept of specific targeting and killing tumor cells using radio labeled antibodies) is limited by the high systemic absorbed dose. Combined e1ectro-radio-therapy could thus be applied to locally treat tumors by enhanced radionuclide uptake and metastases with radioimmuno-therapy.

A further advantage in applying electric pulsation for tumor therapy is the induced enhancement of radiation sensitivity. The enhanced radiation sensitivity might also be applied by using electropulsation combined with either brachy-therapy or external radiationtherapy for the treatment of radio resistant tumors.

During the past two decades, the electropermeabilization technique has developed from being an in vitro tool for laboratory use. Now it can be applied in vivo and used for clinical cancer treatment thanks to enhanced local cytotoxic effect and potential systemic effects by stimulating the immune system.

One of the most promising future use in vivo electropermeabilization is the possibility of gene transfection and using gene therapy in tumor treatment. Gene transfection by electropermeabilization of murine melanoma and C6 glioma has been reported by Rols et aI., (1998) and Nishi et aI., (1996). To promote an immune reaction, the tissue can also be transfected with genes coding for production of molecules that stimulate the immunesystem, such as interferon-y, interleukin-12 or interleukin-18.

REFERENCES

Belehradek, 1., Orlowski, S., Poddevin, B., Paoletti, C. and Mir, L.M., 1991. Electrochemotherapyof

spontaneous mammary tumours in mice. European Journal of Cancer 27,73-76.

Belehradek, M., Domenge, c., Luboinski, B., Orlowski, S., Belehradek, 1. and Mie, L.M., 1993.

Electrochemotherapy, a new antitumor treatment. First clinical phase I-II trial. Cancer 72, 3694-3700.

Biedinger, U., Youngman, R.J. and Schnabl, H., 1990. Differential effects of electrofusion and electropermeabilization parameters on membrane integrity of plant protoplasts. Planta 180,598-602.

143

Ceberg, C.P., Persson, A., Brun, A., Jergil, B., Persson, B.R.R and Salford, L.G., 1992. Macroscopic and

microscopic biodistribution of BSH in a rat glioma model. In: Anonymous Towards Clinical Trials of Glioma with BNCT, pp. 175-178. New York: Plenum Press.

Ceberg, C.P., Brun, A., Mir, L.M., Persson, B.RR and Salford, L.G., 1994. Enhanced boron uptake in RG 2 rat gliomas by electropermeabilization in vivo - a new possibility in boron neutron capture therapy. Anti-Cancer Drugs 5, 463-466.

Chemomordik, L.Y. and Chizmadzhev, YA, 1989. Electrical breakdown oflipid bilayer membranes. Phenomenology and mechanism. In: (Neumann, E., Sowers, A.E. and Jordan, CA Eds.) Electroporation and Electrofusion in Cell Biology, pp. 83-95. New York: Plenum Press.

Cole, KS., 1972. Membranes, Ions and Impulses, Berkeley: University ofCalifomia Press. Cole, KS. and Cole, H.R, 1941. Dispersion and absorption in dielectrics. J. Chem. Phys. 9, 341. Cole, KS. and Guttman, RM., 1942. Electrical impedance of frog egg. J. Gen. Physiol. 25,765.

Coster, H. G.L. and Zimmermann, U., 1975. The mechanism of electric breakdown in membranes of Valonia utricularis. J. Membr. Bioi. 22, 73-90.

Danfeiter, M., Engstrom, P., Persson, B.RR and Salford, L.G., 1998. Effect of high voltage pulses on survival of Chinese hamster V79 lung fibroblast cells. BES conference Aarhus May 23-29 1998 (Abstract).

Dev, S.B., 1994. Killing cancer cells with a combination of pulsed electric fields and chemotherapeutic agents. Cancer watch 3, 12-14.

Dev, S.B. and Hofmann, GA, 1994. Electrochemotherapy--a novel method of cancer treatment. Cancer Treat. Rev. 20, 105-115.

Engstrom, P.E., Salford, L.G. and Persson, B.RR, 1998. Dynamic gamma camera studies of lllIn_

Bleomycin complex in normal and glioma bearing rats after in vivo electropermeabilization using exponential high-voltage pulses. Bioelectrochemistry and Bioenergetics 46,241-248.

Engstrom, P.E., Persson, B.RR and Salford, L.G., 1999a. Studies of in vivo electropermeabilization by gamma-camera measurements of !/9m.yc-DTPA. Biochim. Biophys. Acta (in press).

Engstrom, P.E., Persson, B.RR and Salford, L.G., 1999b. Impedance spectrometry and gamma camera

measurements of !/9m.yc-DTP A for dosimetry and optimization of invivo electropermeabilization. Physics in medicine and Biology (submitted).

Engstrom, P.E., Persson, B.RR, Brun, A. and Salford, L.G., 1999c. A new antitumour treatment

combining radiation and electric pulses. Anticancer Research (submitted).

Engstrom,P.E., 1999d. Electropermeabilization in Experimental Tumour Treatment - Dosimetry and

Tissue Effects. PhD thesis, Radiation Physics, Lund Univeristy, S-22185 Lund, Sweden. Engstrom, P.E., Ivarsson, K, Tranberg, K-G., Stenram, U., Salford, L.G. and Persson, RB.R., 199ge.

Electrically mediated drug delivery for treatment of an adenocarcinoma transplanted into rat liver. Paper VI in PhD thesis, (EngstrOm,P.E., 1999d) Radiation Physics, Lund Univeristy, S-22185 Lund, Sweden

Gabriel, B. and Teissie, J., 1994. Generation of reactive -oxygen species induced by electropermeabilization

of Chinese hamster ovary cells and their consequence on cell viability. Eur. J. Biochem. 223, 25-33.

Gehl, J., Skovsgaard, T. and Mir, L.M., 1998. Enhancement of cytotoxicity by electropermeabilization: an improved method for screening drugs. Anti-Cancer Drugs 9, 319-325.

Glass, L.F., Pepine, M.L., Fenske, NA, Jaroszeski, M., Reintgen, D.S. and Heller, R, 1996. Bleomycin

mediated electrochemotherapy of metastatic melanoma. Arch. Dermatol. 132, 1353-1357.

Glass, L.F., Jaroszeski, M., Gilbert, R, Reintgen, D.S. and Heller, R, 1997. Intralesional bleomycin

mediated electrochemotherapy in 20 patients with basal cell carcinoma. Journal of the American Academy of Dermatology 37, 596-599.

Heller, R, 1995. Treatment of cutaneous nodules using electrochemotherapy. J. Fla. Med. Assoc. 82, 147-150.

Heller, R, Jaroszeski, M., Leo-Messina, J., Perrott, R, Van Voorhis, N., Reintgen, D., and Gilbert, R, 1995. Treatment of B 16 Melanoma with combination of electroporation and chemotherapy. Bioelectrochem. Bioenerget. 36,83-87.

Heller, R, Jaroszeski, M.J., Glass, L.F., Messina, J.L., Rapaport, D.P., DeConti, RC., Fenske, NA, Gilbert, RA., Mir, L.M. and Reintgen, D.S., 1996. Phase 1111 trial for the treatment of cutaneous

and subcutaneous tumors using electrochemotherapy. Cancer 77,964-971.

144

Hibino, M., Shigemori, M., Itoh, H., Nagayama, K. and Kinosita, 1.K., 1991. Membrane conductance of an e1ectroporated cell analyzed by submicrosecond imaging of transmembrane potential. Biophys. J. 59,209-220.

Hou, D. Y., Hoch, H., Johnston, G. S., Tsou, K. C., Jones, A. E., Farkas, R 1., Miller, E. E., and Larson, S. M., 1985a. A new 11 lIn-bleomycin complex for combined radiotherapy and chemotherapy. J.Surg.Oncol. 29,91-98.

Hou, D. Y., Hoch, H., Johnston, G. S., Tsou, K. C., Farkas, R 1., and Miller, E. E., 1985b. Use oflllInbleomycin for combining radiotherapy and chemotherapy on glioma-bearing mice. J.Surg.Oncol. 29, 71-77.

Howell, R W., 1992. Radiation spectra for Auger-electron emitting radionuclides: Report No.2 of AAPM Nuclear Medicine Task Group No.6'). Medical Physics 6, 1371-1383.

Howell, R W., Kassis, A. I., Adelstein, S. J., Rao, D. v., Wright, H. A., Harmn, R N., Turner, 1. E., and Sastry, K. S., 1994. Radiotoxicity of platinum-I 95m-Iabeled trans-platinum (II) in mammalian cells. Radiat.Res. 140,55-62.

Humm, L. H., Rao, D. v., and Howell, R W., 1994. Dosimetry of Auger-electron emitting radionuclides: Report No.3 of AAPM Nuclear Medicine Task Group No.6. Medical PhysiCS 21,1901-1914.

Jaroszeski, M.J., Gilbert, R, Perrott, R and Heller, R, 1996. Enhanced effects of multiple treatment electrochemotherapy. Melanoma. Res. 427-433.

Jaroszeski, M.J., Gilbert, R and Heller, R, 1997a. Electrochemotherapy - An Emerging Drug Delivery Method for the Treatment of Cancer [Review]. Advanced Drug Delivery Reviews 26, 185-197.

Jaroszeski, M.J., Gilbert, RA. and Heller, R,1997b. In vivo antitumor effects of electrochemotherapy in a hepatoma model. Biochimica et Biophysica Acta - General Subjects 1334, 15-18.

Kairemo, K. 1., Ramsay, H. A., Paavonen, T., and Bondestam, S., 1996. Imaging and staging of head and neck cancer using a low pH In-Ill-bleomycin complex. Eur.J. Cancer B Oral Oncol. 32B, 311-321.

Kairemo, K. J., Ramsay, H. A., Paavonen, T. K., 1aaskela Saari, H. A., Tagesson,. M., Ljunggren, K., and Strand, S. E., 1997. Biokinetics of indium-l I I-bleomycin complex in head and neck cancer-implementations for radiochemotherapy. Cancer Detect.Prev. 21,83-90.

Kinosita, 1.K. and Tsong, T.Y., 1979. Voltage-induced conductance of human erythrocyte membranes. Biochim. Biophys. Acta 554, 1015-1019.

Krassowska, W., 1995. Effects ofElectroporation on Transmembrane Potential Induced by Defibrillation Shocks. Pacing and Clinical Electrophysiology (PACE) 19, 1644-1660.

Maccarrone, M., Bladergroen, M.R, Rosato, N. and Finazzi-Agro, A., 1995a. Role of lipid peroxidation in electroporation-induced cell permeability. Biochem. Biophys. Res. Commun 17, 417-425.

Maccarrone, M., Rosato, N. and Finazzi-Agro, A., 1995b. Electroporation enhances cell membrane peroxidation and luminescence. Biochem. Biophys. Res. Commun 206, 238-245.

McLean, 1. R., Blakey, D. H., Douglas, G. R., and Bayley, 1., 1989a. The Auger electron dosimetry of indium-Ill in mammalian cells in vitro. Radiat.Res. 119,205-218.

McLean, 1. R. and Wilkinson, D., 1989b. The radiation dose to cells in vitro from intracellular indium- 111. Biochem.Cell Bioi. 67,661-665.

Mir, L.M., Banoun, H. and Paoletti, C., 1988. Introduction of definite amounts ofnonpermeant molecules into living cells after electropermeabilization: Direct access to cytosol. Exptl. Cell Res. 175, 15-25.

Mir, L.M., Orlowski, S., Belehradek, J. and Paoletti, C., 1991. Electrochemotherapy potentiation of antitumor effect by local electric pulses. European Journal of Cancer 27,68-72.

Mir, L.M .. , Orlowski, S., 1999. Mechanisms of electrochemotherapy. Adv. Drug Delivery Rev. 35, 107-118. Nanda, G.S., Sun, F.X., Hofmann, G.A., Hoffman, RM. andDev, S.B., 1998a. ElectroporationEnhances

Therapeutic Efficacy of Anticancer Drugs: Treatment of Human Pancreatic Tumors in Animal model. Anticancer Research 18, 1361-1366.

Nanda, G.S., Sun, FX, Hofmann, G.A., Hoffman, RM. and Dev, S.B., 1998b. Electroporation Therapy of Human Larynx Tumors HEp-2 Implanted in Nude Mice. Anticancer Research 18, 999-1004.

Neumann, E.M., Sowers, A.E. and 10rdan, C.A., 1989. Electroporation and Electrofusion in Cell Biology,

New York: Plenum Press. Nishi, T., Yoshizato, K., Yamashiro, S., Takeshima, H., Sato, K., Hamada, K., Kitamura, I., Yoshimura,

T., Saya, H., Kuratsu, J., and Ushio, Y., 1996. High-efficiency in vivo gene transfer using intraarterial plasmid DNA injection following in vivo electroporation. Cancer Res. 56, 1050-1055.

Okino. M. and Mohri, H., 1987. Effects of a high-voltage electrical impulse and an anticancer drug on in

145

vivo growing tumors. Jpn. J. Cancer Res. 78, 1319-1321. Ono, K., Kinashi, Y., Masunga, S.-1., Suzuki, M. and Takagaki, M., 1998. Electroporation increases the

effect of Borocaptate, eOS-BSH) in neutron capture therapy. Int. J. Radiation Oncology Bioi. Phys. 42, 823-826.

Panje, W.R, Hier, M.P., Garman, G.R, Harrell, E.H., Goldman, A. and Bloch, 1., 1998. ElectrOporation Therapy of Head and Neck Cancer. Ann. Otology, Rhinology & Laryngology 107, 779-785.

Poddevin, B., Orlowski, S., Belehradek, J.J. and Mir, L.M., 1991. Very high cytotoxicity of bleomycin

introduced into the cytosol of cells in culture. Biochem. Pharmacol. 11,67-75. Rols, M. P., Delteil, C., Golzio, M., Dumond, P., Cros, S., and Teissie, 1., 1998. In vivo electrically

mediated protein and gene transfer in murine melanoma. Nat.Biotechnol. 16, 168-171. Rudolf, Z., Stabuc, B., Cemazar, M., Miklavcic, D., Vodovnik, L. and Sersa, G., 1995.

Electrochemotherapy with bleomycin. The first clinical experience in malignant melanoma patients. Radiol. Oncol. 29,229-235.

Sale, A.J.R. and Hamilton, W.A., 1968. Effects of high electric fields on microorganisms. III. Lysis of

erytrocytes and protoplasts. Biochem. Biophys Acta 163, 37-43. Salford, L.G., Persson, RB.R., Brun, A., Ceberg, C.P., Kongstad, P.C. and Mir, L.M., 1993. A new brain

tumour therapy combining bleomycin with in vivo electropermeabilization. Biochem. Biophys. Res. Commun 194, 938-943.

Sersa, G., Cemazar, M. and Miklavcic, 0.,1995. Antitumor effectiveness ofe1ectrochemotherapywith cisDiamrninedichloroplatinum(II) in mice. Cancer Research 55, 3450-3455.

Sersa, G., Cemazar, M., Menart, v., Gabercporekar, V. and Miklavcic, D., 1997a. Anti-Tumor Effectiveness of Electrochemotherapy with Bleomycin is Increased by TNF-alpha on SA-1 Tumors

in Mice. Cancer Letters 116, 85-92 .. Sersa, G., Miklavcic, D., Cemazar, M., Belehradek, 1., Jarm, T. and Mir, L.M., 1 997b.

Electrochemotherapy with COOP on LPB Sarcoma - Comparison of the Anti-Tumor Effectiveness in Immunocompetent and Immunodeficient Mice. Bioelectrochemistry & Bioenergetics 43, 279-283.

Shibata, Y., Matsumura, A., Yamamoto, T., Nakagawa, K., Yoshii, Y., Nose, T., Sakata, 1., Nakajima, S., Hayakawa, Y. and Ono, K., 1998. Neutron Capture Therapy with a New Boron-Porphyrin Compound in the rat 9L Glioma Model. J. Exp. Clin. Cancer Res. 17,285-289.

Siesjo, P., Visse, P., Lindvall, M., Salford, L.G. and Sjogren, H.-D., 1993. Immunization with mutagentreated (tum-) cells causes rejection ofnonimmunogenic rat glioma isografts. Cancer Immunol

Immunother 37,67-74. Stampfli, R., 1958. Reversible electrical breakdown of the excitable membrane of a Ranvier node. Ann.

Acad. Bras. Cien. 30, 57-63. Suzuki, T., Shin, B.C., Fujikura, K., Matsuzaki, T., Takata, K., 1998. Direct gene transfer into rat liver

cells by in vivo electroporation. FEBS Letters 425, 436-440. Tien, H.T., 1974. Bilayer Lipid membranes (BlM): Theory and Practice, New York: Marcel Dekker, Inc.

Tomov, T.C., 1995. Quantitative dependence of eIectroporation on pulse parameters. Bioelectrochemistry

and Bioenergetics 37, 101-107. Tounekti, 0., Pron, G., Belehradek, 1.1. and Mir, L.M., 1993. Bleomycin, an apoptosis-mimetic drug that

induces two types of cell death depending of the number of molecules internalized. Cancer

Research 53, 5462-5469. Wratten, M.L., Van Ginkel, G., Van't Veld, A.A., Bekker, A., Van Fassen, E.E. and Sevanian, A., 1992.

Biochemistry 31, 10901-10907.

Xie, T.D., Sun, L., Zhao, H.G., Fuchs, J.A. and Tsong, T.Y., 1992. Study of mechanisms of electric fieldinduced DNA transfection IV. Biophys. J. 63, 1026-1031.

Zimmermann, U., Pilwat, G. and Riemann, F., 1974. Dielectric breakdown of cell membranes. Biophys. J.

14, 881-899. Zimmermann, U., Scheurich, P., Pilwat, G. and Benz, R., 1981. Cells with manipulated functions: New

perspectives for cell biology, medicine and technology. Angew. Chem. Int. Ed. Engl. 20, 325-344.

Zimmermann, U., 1982. Electric field-mediated fusion and related electrical phenomena. Biochem. Biophys Acta 694, 227-277.

Zimmermann, U. and Neil, G.A., 1996. The effect of high intensity electric field pulses on eukaryotic cell membranes: fundamentals and applications. In: Anonymous Electromanipulation of cel/s, pp. 4-

80. CRC Press.

146

THE ELECTRIC FIELD-INDUCED ELECTRO CONFORMATIONAL COUPLING OF CELL MEMBRANE PROTEINS

Wei Chen

Departments of Dermatology, Bioengineering, and Physiology and Biophysics The University of Illinois School of Medicine Chicago, IL 60612

I. INTRODUCTION

Since its discovery in the later nineteenth century, electricity has become the cleanest, most efficient and most widely used energy form in our daily life. Because of increasingly demanding for electrical energy in the home and workplace, it seems like that trauma from electrical shock continue to increase. Better understanding of the underlying mechanisms of electrical injury and the development of efficient therapeutic techniques and management procedures for electrically injured patients are the common goals of medical doctors, engineers and scientists. On the other hand, there are many proteins in our body, especial those in cell membranes, are either voltage-dependent or electrigenic. Those proteins are mainly responsible for the cellular electrical activities. How those membrane proteins interacted with an applied electric field is a fascinating question, which has drawn more and more attentions.

It is estimated that 4% of all the patients admitted to hospital bum units in the United States are electrical injury patients (DiVincenti et al., 1969). Most accidents are related to electrical potentials greater than 1000 volts (Hunt, 1992). Partially because of incomplete understanding of the pathophysiology of tissue damages and difficulties in accurately diagnosing and treating electrically injured patients, mortality rate for accidentally electrical injury ranges from 3 to 15%; annually, more than 1000 people in the United States alone die because of electrical shock-related injuries (Gourbiere, 1992).

Over the last two decades, there has been only incremental improvement in the understanding of underlying mechanisms and the pathophisiology of electrical injury. Clinical

Advances in Electromagnetic Fields in Living Systems, Vol. 3 Edited by J.c. Lin, Kluwer AcadennclPlenum Publishers, 2000 147

outcomes for victims of electrical trauma have also improved, but only marginally. Joule heating has long been recognized as a primary cause of tissue damages (DiVincenti et al., 1969; Rouge and Dimic, 1978; Chibert et al., 1985; Bingham, 1986; Hunt, 1992). Chilbert et al. (1985) passed electrical currents through the hind limbs of dogs until the temperature of the gracilis muscle reached 60°C and found a strongly increase of cells damage and decrease in electrical impedance (Chibert et al., 1985). However, Tropea and Lee (1992) recently computer-simulated a model of a 10 kV hand-to-hand shock using worst-case conditions such as perfect contact with electrodes. The numerical analysis results showed that after an exposure to a 10 kV shock for one second, the temperature of the medial tissue of the human forearm did not reach 42°C, the critical temperature for thermal tissue damage. Some ofthe tissue damages seen in electrical shock injury must thus result form a non-thermal mechanism that may be related to the electric field-induced supramembrane potential difference. One of the hypotheses for cell membrane damages in electrical injury is membrane electroporation. A growing body of evidence indicates that electromediated pores or pore-like structures form on the phospholipid bilayer of cell membrane during exposure to high voltage electrical fields, resulting in leakage of ions and cellular contents out of the cells (Lee and Kolodncy, 1987; Lee et al., 1988). These cells and tissues may later display edema and finally undergo necrosis because of the loss of cellular metabolites.

In the normal physiological situations, proteins often undergo large structural changes (conformational transitions) in carrying out their biological functions. An engineer term, "dielectric constant change" has been widely used to macroscopically describe the ability of the protein to be affected by an applied electric field. These proteins' conformational changes inevitably results in significant changes in the electrical properties of the proteins. This electrical susceptibility strongly suggests that membrane proteins may well be structurally damaged and functionally disabled during the exposure to an high-voltage electrical field. In the clinic, we often see electrically injured pa~ients, whose skin and muscles around the current pathway show no sign of damage. However, the functions of the injured limbs or organs show significantly reductions. These observations imply that in addition to membrane e1ectroporation, proteins, especially the membrane proteins, may suffer some structural and functional damages during the electric injury.

A typical example is ion channel, such as the voltage-dependent Na+ and K+ channels, which are the key molecules necessary for generating an action potential and maintaining the membrane resting potential. Recently, an in vivo study on rat skeletal muscles showed that a highvoltage electrical shock may significantly decrease the magnitude, and increase the latent period of the membrane action potential. These results indicate that the membrane proteins of ion channels may suffer damage during the electrical shock. Moreover, Poloxamar, a surfactant chemical has been shown to have therapeutic sealing effects on the permanent electropores of the cell membrane (Lee et al., 1992; Chen and Lee, 1992; Sharma et al., 1996). Therefore, it becomes more and more important to understand the more subtle mechanisms involved in the electrical injury damaging the membrane proteins.

Either the membrane e1ectroporation causing a loss of ionic concentration gradient or the damage to membrane proteins resulting in dysfunctions of ion channels, a common result is that the cells lose their physiological membrane potentials, including depolarization of the membrane resting potential, and reduction of the membrane action potential. A clinical example is the side

effects from defibrillation treatments for cardiac disease patients. The use of a controlled electrical field to manage cardiac arrhythmia is a common practice today. When using a controlled external electrical field to manage arrhythmia of ventricular muscle, one of the side effects is a reduction of the membrane resting potential. Strong defibrillation shocks can also prolong the membrane

148

depolarization and prevent cellular activation for seconds or minutes. Of even greater interest, depolarization of the membrane resting potential is a common

characteristic in many diseases. In addition to the side effects of defibrillation treatment (John et 01., 1987), a short list also includes electrically injured skeletal muscle fibers (Chen and Lee, 1994, b) and nerve cells (Abramov et 01., 1995), cardiac stroke (Downar et 01., 1977), coronary artery disease (Kiline and Fozzard, 1984; Ursell et 01., 1985; Tseng et 01., 1987), hypoxic pulmonary blood vessel (Gin, 1995) and epilepsy (Stasheff et 01., 1992).

On the other hand, many cells or organisms are able to detect both gradual and abrupt changes in their environment in order to survive. These cells are also able to coordinate with other cells, organs or organisms in order to engage in, and regulate their wide varieties of activities. For examples, sharks, skates and rays are able to perceive electric signals in an order of magnitude only ofnano-voltage per centimeter (1); electric fishes use very weak electric organ discharge (BDO) to warn or track their preys (2); and birds navigate according to the earth magnetic fields.

More specifically, many molecular pumps on cell membranes work to maintain, at the expenditure of energy, differences in electrochemical potential of ions (H+, Na +, K+, Ca2+) between cytoplasm and extracellular medium. Among these electrogenic pump molecules, the Na+fI(+ ATPase molecules have the largest population in skeletal muscle fibers and playa significant role in membrane electrophysiological activities. Since 3 Na+ ions are extruded for every 2 K+ ions pumped in across the cell membrane, each pump cycle results in a net loss of a positive charge. Thus, electrogenic pump currents have two effects on the membrane resting potential: electrogenic potential and diffusion potential. These electrogenic pump currents have important electrophysiological effects by driving the membrane potential away from its channel-open threshold potential, hence, the cells become quiescent. Electrogenic pump currents also contribute to repolarizing the cell membrane during the falling phase of the action potential. The Na+fI(+ pump currents that hyperpolarize the cell membrane accelerate the return of the depolarized cell membrane to the resting potential.

How to efficiently restore the electrical shock-reduced membrane resting potential and to maintain the cellular functions for electrically injured patients are fascinating questions. One of the promising attempts is to apply a weak, oscillating electrical field to the injured tissues in order to activate those electrogenic, molecular pumps. This is based on a fact that almost all of the electrogenic, molecular pumps in cell membrane are sensitive to changes in the membrane potential difference. Therefore, the changes in membrane potential induced by an applied electric field should affect the functions of the electrogenic pump molecules.

This turns to the second part of this review, in use of an oscillating electric field to facilitate the functions of the membrane proteins. From the viewpoint of thermodynamic basis, many types of oscillating perturbation can actively affect the membrane transporters. Some scale perturbations, such as temperature, ion concentration and pressure may interact with molecules to generate a vectorial reaction when the molecules are exposed to an anisotropic medium. Exposure to an anisotropic medium and performing an anisotropic reaction are essential factors for the molecules absorbing energy from an oscillating perturbation. Membrane bounded enzymes are typically in this category of proteins. Among different types of perturbation, electric field is particular interesting because it is vectorial characteristics.

Experimental studies from several laboratories have demonstrated that certain cells can transmit information through membrane enzymes and absorb energy from an oscillation perturbation. For example, cells can transmit information through oscillating Ca2+ and cAMP concentration, as in p-cells of pancreatic islets (Atwater et 01., 1978, 1979; Chay and Keizer,

149

1983; Lechleiter et al., 1991) and in slime molds (Tomchik and Devreotes, 1981). These procedures may result from the membrane enzymes or signal receptors receiving, processing and transmitting energy from the oscillating electrical potential or chemical potential. These periodical signals couple with the membrane proteins by influencing their rate constants and equilibrium constant of the proteins' catalytic reactions.

Different theories have been postulated to interpret the oscillating electric field-induced activation of the membrane active transporters. One is electroconformational coupling (BCC) considering transduction of electrical energy into membrane enzyme and receptors. This theory has been successfully used to interpret the electric field-induced cation pumping activities of the Na+/K+ ATPase and the Ca2+ ATPase of human erythrocytes, and the ATP synthetic activity of beef heart mitochondrial ATPase. The second theory is about low periodic electric field induced effects. Several models have been proposed. Blank (1987) set up a surface compartmental model (SCM) considering the current produced in the vicinity of the cell membrane as the origin of the field effect. This theory has been successfully used to interpret AC stimulation and inhibition of the ATP hydrolysis activity of the Na+/K+ ATPase and the generation of frequency optima (Liu et al., 1991). Markin et al. (1992) have also postulated an oscillatory activation barrier model (OAB), which considers resonance transduction between an oscillating field and the activation barrier of the rate limiting step in an enzymic reaction. This model has successfully explain the data of the AC stimulated ATP hydrolysis activity ofEcto-ATPase from chicken oviduct and FoFI-ATPase from beef heart.

The main difference between the theory of electroconformational coupling and the theory oflow field activation, either SCM and OAB, is the field-strength or the field-induced membrane potential difference. When the apparent field-strength inside cell membrane or the external electric field-induced membrane potential difference is comparable with the physiological value, the ECC model is easy to understand. On the other hand, experimental results have been shown that the induced field-strength inside the cell membrane lower than the physiological value, such as a membrane potential difference in a few 11 V, may also have effects on the cell functions. This extremely low field may stimulate enzyme activity, inhibit synthesis of DNA, RNA and proteins, and induce many other cellular responses. The low field activation theories will be helpful to interpret these experimental data.

In this article, I will review the results of theoretical and experimental studies conducted by others as well as by our own, in an intensive electric field-induced conformational damages to the membrane proteins; and in a weak, oscillating electric field-coupled conformational changes and activations of the membrane active transporters. The low field activation theories which does not involve proteins conformational changes, are out of the scope of this review and will only be briefly described.

n. CELL MEMBRANE: THE TARGET OF ELECTRIC FIELD

1. Membrane Structures and Membrane Potentials

1) Intrinsic Membrane Potential

Membrane potential can be simply viewed as a linear membrane potential drop across the membrane. There are mainly two types of intrinsic membrane potentials pertinent to membrane

150

structure and chemical potential: transmembrane potentials and surface potential. The two potentials are thermodynamic equilibrated which influence the molecular free energies as well as the dynamics of the conformational transitions of many membrane proteins and the phospholipid bilayer.

Approximately 10-20% of the phospholipid in a cell membrane bear negative charge. Astumian (1994) estimated the surface charge density and found it can be as high as one electric charge per 300 A2. These negative charges generate a potential difference on the same side of the membrane which can be defined as a potential difference between the surface and the bulk solution. These inside and outside surface potential ofthe cell membrane are denoted as fl "' .... and fl "' .. out, respectively. The surface potential is approximately -60 m V relative to the bulk solution. This surface charge is screened by positive ions in the solution and thus falls to 60/e mV in one Debye distance, which at 0.1 M ionic strength is 1 nm. Hence, the electric field within the Debye distance of the surface is roughly 200 kV/cm, which significantly influence the concentration of the charged particles near the cell membrane. As a result, concentrations of positive ions with single charge near the cell membrane will be lO-fold larger than those far away from the cell membrane in the bulk solution.

O.2nm k= Inm

Figure 1. Schematic representation of the membrane and surface electric potentials in the vicinity of the cell membrane. The relevant potentials are the inside and outside surface potentials of t:.1jTs.in and t:.1jTs.ouu

the diffusion potential t:.ljTdilT, and the bulk-to-bulk potential, t:.ljTb' Taken for Astumian, 1994.

The transmembrane potential is a diffusion potential, which depends on to factors: the ionic concentration gradient across the membrane, and the differential membrane permeabilities to these ions. This transmembrane potential, or the equilibrium potential, flljT diff, can be easily

151

calculated according to the Goldman-Hodgkin-Katz (GHK) equation. Ifthere are more than one kind of ions at the two sides of the membrane, the transmembrane potential is primarily determined by the equilibrium potential of ions to which the membrane has larger permeability. Under physiological solutions, the concomitant membrane resting potential for the skeletal and cardiac muscles is around -90 mV(inside negative). This is equivalent to a field-strength within the membrane of almost 200kVlcm.

The bulk to bulk potential is a combination of the surface potentials, a Ijr ", .. and a Ijr ",out> and the transmembrane potential, aljrdiff. Because of the opposite polarity of the inside and outside surface potential, the bulk-to-bulk potential is numerically equal to the transmembrane potential. In fact, membrane potential we macroscopically measured is the bulk-to-bulk potential. Figure 1 shows a schematic diagram of these membrane potential differences across the cell membrane as well as their relationship.

2) Field-Induced Membrane Potential

When living cells are exposed to a high voltage electric field, the cell membrane will suffer the major electric injury. That is because the electrical resistance ofthe phospholipid bilayer of the cell membrane is six to eight orders in magnitude higher than that of cytoplasmic and extracellular electrolytes. Therefore, when exposure to an external electric field, most of the fieldinduced voltage-drop falls on the cell plasmic membrane.

The external field-induced transmembrane potential, aljr, can be expressed

A 'I' = 1.5RceltEatCOsO[1- exp( -t / Tmem)]

Tmem = RcellCmem(rin + rout 12) (1)

where Eext is the field-strength of the applied electric field and R.eu is the radius of the cells. e is the angle between the field and the normal of the cell membrane, and Cmem, rin and rout are the capacitance per unit area of the membrane and the resistivities of the cytoplasmic fluid and external medium, respectively (Cole, 1972; Kinosita and Tsong, 1977; Neumann et al., 1989). For biological cells of diameters in micrometer, 't'mcm is less than 1 ~s. For cells with larger diameters, the 't'mem will be greater. When the duration of a direct current (DC) electric pulse longer than 't'mem' the induced maximum potential difference across the cell membrane is

A 'I' = 1.5 ReellEat (2)

Because the thickness ofa cell membrane is much smaller than the dimension of the cell, and because the resistivity of the cell membrane is much higher than that of intracellular fluid or extracellular medium, the induced field-strength inside the cell membrane is much larger than the apparent strength of the applied electric field. The equivalent intensity of the field within the cell

membranes is the apparent field-strength amplified by the ratio of cell radius, R.:cll to the membrane thickness, d (Cole, 1972).

EequiOO Eat * Ecell 1 d (3)

152

Excitable cells, such as skeletal muscle fibers and nerve cells, the ratio of the cell size to the membrane thickness is easily on an order of thousands or more. Therefore, the induced equivalent field-strength inside the cell membrane is hundreds and thousands times larger than the apparent strength of the applied electric field (Cole, 1972). Under such a high field strength, both the membrane phospholipid bilayer and the membrane proteins may inevitably undergo structural changes. These structural changes induced by a supramembrane potential differ from the Joule heating effects on the cell membrane which mainly results from a large transmembrane current. These structural changes, if occur in the phospholipid bilayer, is referred to the membrane electroporation; and if occur in the membrane proteins, is so-called membrane proteins conformational damage.

3) Phospholipid Bilayer Versus Membrane Proteins

In order to separate the electric field-induced effects on the membrane proteins, such as ion channels and active transporters, and on the membrane pho pholipid bilayer, it is necessary to distinguish the reference membrane potential with respect to the phospholipid bilayer and the reference potential to the membrane proteins. The phospholipid bilayer, the major component of the cell membrane, is organized with the hydrophobic tails of the phospholipid structure on the inside of the membrane and hydrophilic heads facing outward (Lee and Kolodncy, 1987; Lee et al., 1988). The amphipathic structure of the phospholipid membrane is to maintain the integrity of a hydrophobic permeability barrier in close proximity to an aqueous medium. One of the characteristics of the phospholipid bilayer that differs from the membrane proteins is their symmetric structure. This symmetric feature determines that the electrical field-induced effects on the phospholipid bilayer are independent on the polarity of the applied electric field (Tung, 1992; Chen and Lee, 1994, a).

In contrast, the pattern in which proteins are embedded in the phospholipid bilayer determines the asymmetric structures of the membrane proteins, especially the voltage-gated ion channels and the electrogenic molecular pumps. Almost all of these voltage-gated ion channels and active transporters show ion- and ligand-selectivities. Because of these selectivities and the ionic concentration gradients across the cell membrane, the membrane potential is equilibrated at the membrane resting potential, -90 m V for natural intact fibers of skeletal and cardiac muscles. Therefore, in order to study an external field-induced effects on the membrane proteins, we must account for this resting membrane potential.

2. Voltage-Sensitive Membrane Proteins

Approximately 40% of the cell membrane, by weight, consists of proteins functioning as ionic channels, transporters and signal receptors. The most basic subunits, the amino acids of the membrane protein molecules carry electrical charges. In general, each peptide unit represents an electric dipole moment of about 3.5 Debye (D). For example, in an a-helical structure, which is a major functional structure of the voltage-gated Na+, Ca2+ and K+ channel proteins, many small peptide dipoles are aligned almost perfectly to form larger dipoles with a order of 120 D across the cell membrane (Wada, 1976; Hoi et aI., 1978, 1985). These electric dipoles are strongly affected by an extemal electrical field. In addition, some charged elements are moveable in response to physiological changes in the membrane potential. They are often expressed as gating currents in channel proteins (Armstrong, 1981) and intramembrane charge movement currents in

153

other voltage-dependent membrane proteins, such as electrogenic pumps (DeWeer et al., 1988; Lauger, 1991), and dihyropuridine receptors, or the voltage-sensors in skeletal muscle fibers (Schneider and Chandler, 1973; Adrian and Peres, 1977; Rios and Brum, 1987; Tanabe, et al., 1990). The external electric field-induced supramembrane potential exerts extraordinary force on these moveable charged particles in the membrane proteins, resulting in proteins' conformational changes.

In general, supramembrane potential may cause the charged particles moving along or reorienting the equivalent dipoles in alignment with the direction of the external electrical field. More specifically, the membrane proteins may undergo some of the following processes: 1) the permanent dipole may reorient, or the randomly positioned weak dipole may form new permanent dipoles, to parallel to the direction of the applied electrical field; 2) the ionized side-chain may dissociate from the main amino acid chain or the electrically charged subgroup may separate from binding on the protein; and 3) physiologic moveable charges particles, such as gating particles and intramembrane charge movement particles, may lose their moving capability by sticking on or separate from the neighboring amino acids. These conformational alterations may result in dysfunctions of the membrane proteins.

Current knowledge of the primary structures of the principal subunits of the voltagesensitive ion channels has provided a molecular template for the voltage-dependent membrane proteins. Substantial progress has been made toward the development of a functional map of the alpha-subunit of the channel proteins (Catterall, 1988). One example is the hypothesized channel gating system. This unique structure of the S4 segments of the four inter-repeat domain, consisting of repeated motifs of positively charged amino acid residue followed by two hydrophobic residues, has been suggested as a voltage-dependent gating system with a helical screw model (Catterall, 1988; Guy, 1992). This voltage-dependent gating system is inevitably susceptible to an applied high-voltage electric field.

In addition to ion channels, functions of many other membrane proteins depend on the membrane potential by virtue of their intramembrane charge meovement. One example is the dihydropuridine receptors, the voltage-sensor in the transverse (T) tubular membrane of skeletal muscle fibers, an important step involved in excitation-contraction coupling.

3. Types of Voltage-Sensitive Membrane Proteins

Figure 2 shows different kinds of proteins that are known to be electrically active. The first is the voltage-gated ion channels, such as the voltage-gated Na+, K+ and Ca2+ channels. The functions of these channel proteins depend on the membrane potential because of the existence of physiologically movable charged particles, or voltage sensors, intrinsic to the protein structure. Among these channels, the sodium channel is the most sensitive to the membrane potential. The gating charged particle, or the voltage-sensor, for opening a Na+ channel is equivalent to six elementary charge, comparing to only about four equivalent charged particles in the K+ channels (Hille, 1992; Almer, 1978). Therefore over 90% channels switch to open state from its resting state within a few millivolts change to the membrane potential, resulting in a sharp rising phase of an action potential.

Next, it is a ligand-gated channel. A typically example is the acetylcholine receptor. Indeed, this kind of channels are not necessarily being voltage-gated. However, most of these channels, when open, function as ion channels with a significant increase in ion permeability.

154

Because oflarge ionic current passing through these ligand-gated channels, the open probability of these channels has generally significant dependence on the membrane potential.

3No'

AOP+P, H'

Figure 2. A few examples of electrically active membrane proteins embedded in and affiliated to the cell membranes. Taken from Astumian, 1994.

The following example is electrogenic pump, such as the Na+fI(+ ATPase and Ca2+_ ATPase. From their functions of electrogenecity, it is not difficult to imagine these pump molecules are susceptible to an electric field. For example, the Na+fI(+ ATPase extrude three Na+ ions for every two K+ ion pumping into the cell by expenditure of one ATP molecule (Lauger, 1991). There are two kinds of contributions of the Na+fI(+ ATPase to the membrane potential: electrogenic potential and diffusion potential. The main function of the Na+fI(+ ATPase is to sustain the physiologic ionic concentration gradient across the cell membrane, which is specially crucial in many cellular processes of excitable cells, including maintenance of the membrane resting potential and generation of action potentials.

Special emphasis has been put on in several laboratories to understand the interaction of the Na+/K+ pump with an applied oscillating electric field. Tsong and colleagues (Teissie and Tsong, 1980, 1981) have experimentally studied the activation effects of an oscillating electric field on the Na+/K+ ATPase. They found that an osciIIating field can strongly facilitate the functions of the pump molecules, from which they hypothesized the electrical coupling of the pump's intrinsic moveable charged particle. This result is consistent with the results from DeWeer, Gadsby, and Rakowski (1988), that in the Na+ transition limb two negative charged particles moving across the cell membrane are involved in each pumping cycle. These intrinsic charged particles move in the pump molecules in response to the changes in membrane potential even without binding to ions in the solution.

The next category of potential-dependent membrane proteins includes F of 1 proton ATPase which is typically found in mitochondria in synthesis of A TP molecules. The susceptibility ofthese enzymes to membrane potential is by virtue of the driving force from proton electrochemical gradient (Lauger, 1991). The last proteins, phosphorylase kinase, is included in the Figure just to point out that many signal transduction proteins, even not explicitly regulated by the membrane potential, may tum out to be electrically sensitive, especially at the field-strength involved in electrical injury.

155

4. Directionality of Cell Membranes

1) Directionality of Cell Membrane

Previously, investigators assumed that during exposure to an externa~electrical field the field-induced potential differences across the two hemispheres ofthe cell membranes surrounding the cell were symmetric, with opposite polarity. Recently, however, several investigators found that the extent of electrical field-induced electro po ration on the membranes of the two opposite hemispheres is different (Rossignol et al., 1983; Mehrle et al., 1985; Sowers, 1988; Telke et al., 1990; Masahiro et al., 1991). Some controversy results have also been reported. Kodama et al. investigated the effects of electric field stimulations, at a voltage gradient used in clinical DC defibrillation, on the transmembrane potential and the contraction of isolated guinea pig ventricular muscle (1994, 1998). They found that stimulation pulse with an intensity above 15 Vlcm causes a prolonged depolarization, an oscillation of membrane potential, repetitive spontaneous activity, and an increase in the contractile force. These electric stimulation-induced effects were independent of the phase of shock application in relation to the basal action potential. Kinosita et al. studied asymmetric feature of electroporation in sea urchin eggs and lip somes (Masahiro, 1991). In the case of sea urchin eggs, higher permeability was observed on the hemisphere facing the negative electrode, while the positive-electrode side was more permeable in liposomes. With few exceptions (Dimitrov and Sower, 1990), these investigators reported that any membrane area of a cell which is closer to a positive electrode will show more evidence of electro po ration than a membrane area closer to a negative electrode.

2) Protection Effects on Cell Membrane

Several hypotheses have been developed to explain these phenomena. Since a large potential difference across the cell membrane is considered as the cause of electro po ration, some investigators (Telke et al., 1990) have suggested that the physiologic resting membrane potential, which is vectorially added to the external field-induced membrane potential, results in asymmetrical electroporation on the spatially opposed membranes. Sowers (1988) has postulated that the phenomenon of electro-osmosis may explain asymmetrical transport. It was also shown that the asymmetric membrane electroporation of the two opposite hemispheres could be due to an artifactual process (Sowers, 1988; Dimitrov and Sowers, 1990). Most ofthese studies involved using optical dye measurements of non-excitable cells, such as nucleated plant, liposomes or erythrocyte ghosts. The phenomena of asymmetrical membrane electroporation were observed only after the application of high-voltage electrical shocks.

Recently, by utilizing the improved double Vaseline-gap voltage clamp technique, we investigated asymmetrical electro-mediated permeabilization in cell membranes of frog skeletal muscle fibers during exposure to high-voltage electric shocks (Chen and Lee, 1994, c). A negative shock pulse that hyperpolarizes cell membrane introduces much larger transmembrane leakage current than a positive shock pulse that depolarizes the cell membrane. Our hypothesis is that in the situation of skeletal muscle fibers the membrane proteins, especially ion-channels, play a significant role in differential electroporation in the opposite membranes. When exposed to a highvoltage electric field, the increase in membrane conductance due to ion channel-opening on the depolarized hemisphere initiates a redistribution of the membrane potential differences across the two opposed membrane hemispheres.

156

To test our hypothesis, we have perfonned a series of experiments to clarifY the asymmetrical membrane electro-mediated penneabllization during a high-voltage electrical shock by distinguishing the voltage-gated ion-channel currents and the membrane electroporated currents. The purpose of the studies is to identify the changes in the field-induced membrane potential due to the channel-opening, and to investigate their results in protecting the cell membrane from electroporation.

Channel-mediated transmembrane currents in response to a variety of high-voltage electrical pulses were resolved from the pulse-induced electroporated leakage currents. Under the voltage clamp condition, the electropore-mediated current as Ii percentage of the total transmembrane currents has been quantified. This percentage starts from zero at a low membrane potential and then increases with the membrane potential increases.

I ··0.6

0.5

1 0.4

8.

! O.l

'a 0.2

fo S

~ 0.1

~ 0.0 0 120 240 l60 ·720

Shock pulses (mV)

Figure 3. The relationship between the magnitude of positive stimulation pulses and the percentage of the pulse-induced electroporated currents with respect to the total transmembrane current response. Taken from Chen et 01., 1994, c.

Figure 3 shows the percentages of electr-oporation currents with respect to the total transmembrane currents corresponding to positive pulses with varying magnitudes. This percentage remains zero when the stimulation pulse is below the threshold for membrane electroporation. It then increases with an increase in the pulse magnitude, but never reaches 100010 except when all of the protein channels are damaged.

This result shows that the voltage-gated ion channels provide extra pathways for the fieldinduced transmembrane current, that plays a salutary role in protecting the membrane phospholipid bilayer from electroporated. It is important to notice that these results were obtained when using a voltage clamp, where the membrane potential was constrained regardless of changes in the membrane conductances.

At a low membrane potential, the positive pulse induced membrane conductance is much higher than a negative pulse induced membrane conductance. When the electrical pulse increased. this difference is gradually decreased. Figure 4 shows the ratio of membrane conductances of the depolarized cell membrane to the hyperpolarized membrane as a function of the magnitude of

157

electrical pulses. At a low pulse magnitude, the membrane conductance of the depolarized membrane is 2 to 3 times that of the hyperpolarized membrane.

3.5

~ 30 5

0 ::I

'" C 2.5 0 0 ., ~ 20

~ ., e 1.5 'a 0 '\.,: .= 10 N ~

05 0 120 240 lGO 480 600 720

Shock pulses (mV)

Figure 4. The ratio of the membrane conductance in response to positive stimulation pulses relative to that in response to negative pulses as a function of pulse magnitude. Taken from Chen et al., 1994, c.

Channel-opening in the depolarized cell membranes can generate 2 to 3 times of membrane conductances of the hyperpolarized membrane in response to the same membrane potential. Indeed, our experiments were conducted under a voltage clamp. However, our results implies that open of ion-channels can generate a significant difference in the membrane conductance in response to different polarities of an electric field.

In the real situation, two membrane hemispheres surrounding the cell are electrically connected in series with respect to an external electrical field. The field-induced total potential difference across the whole cell is constant as long as the size and shape of the cell remain the same. Distribution of the membrane potentials across the two opposed hemispheres depends upon their membrane conductances. The opening of the ion-channels in the membrane facing the cathode causes an increase in membrane conductance and, consequently, resulting in a reduction in the field-induced transmembrane potential. The membrane potential at the hemisphere facing the anode will spontaneously increase as a consequence. As a result, less electroporation occurs on the depolarized membrane hemisphere facing the negative electrode.

These results indicate that the voltage-gated ion channels have a protective effect, mediating some of the damage to the phospholipid membrane from electroporation by adjusting the membrane conductance.

3) Membrane Resting Potential and Phospholipid Bilayer

The membrane resting potential, as a constant, is vectorially added to the external fieldinduced membrane potential, which also results in unequal membrane potentials across the two opposite membrane hemispheres. The polarity of the field-induced membrane potentials across the two membrane hemispheres is always opposite, however, the magnitude is not necessarily the same. Distribution of the field-induced membrane potential across the two hemispheres depends

158

on the membrane conductances of the two hemispheres. The higher the membrane conductance, the lower the field-induced membrane potential is. In fact, the distribution of the membrane potentials across the two membrane hemispheres involves a dynamic procedure of potential redistribution depending upon changes in the membrane conductance. This redistribution of membrane potential does not involve the membrane resting potential.

Assuming that the membrane resting potential is zero, the membrane hemisphere with open voltage-gated ion channels has a larger membrane conductance than the hemisphere with fewer open channels. The difference in the membrane conductances results in a difference of the field-induced voltage drop across the two membrane hemispheres. The consequent result is that the hyperpolarizing membrane potential is much higher than the depolarizing potential. Meanwhile, because of continuity of electrical current, the current crossing the two membrane hemispheres must be the same. The influx current passing the hyperpolarized membrane hemisphere, where no channel opens, must be equal to the efflux current passing the depolarized hemisphere, where many channels open. Consequently, the higher membrane potential on the hyperpolarized membrane hemisphere with fewer open ion channels results in a larger formation of electrically-mediated pores.

In other words, our theory of the protection effects of ion-channels on cell membrane from electroporation is independent of the membrane resting potential.

m SUPRAMEMBRANE POTENTIAL-INDUCED ELECTRO CONFORMATIONAL DAMAGES IN MEMBRANE PROTEINS

1. Membrane Electroporation

When a strong electrical field is applied to cell membrane, electroporation occurs because the electric field disrupts the structural integrity of the cell membrane (Tsong, 1991). One consequent result of the membrane electroporation is a dramatic increase in the non-selective permeability and conductivity of the cell membrane, resulting in exchange of contents between the cell and its extracellular environment.

Today, electroporation technique has been widely used in molecular biology research and clinic practice in order to facilitate uptake or discharge of a variety of molecules in living cells (Zimmermann eta I., 1980; Wang and Neumann, 1982; Winegar eta I., 1989; Chang eta I., 1991). Typical examples include the electric field-induced cell fusion and gene transfection. In general, relatively small, non-excitable cells have been used. Recently, Mir et al. suggested that electroporation technique can be applied to the stratum corneum of skin for the purpose of percutaneous drug delivery (Mir et al., 1991). Weaver and his colleagues have demonstrated from in vivo studies that a few microsecond duration pulse of an intensity electric field can significantly increase permeability of the stratum corneum up to hundreds and thousands times (prausnitz et al., 1993).

On the other hand, the use of controlled electrical fields to manage cardiac arrhythmia is a common practice today. How to increase the defibrillation efficience and decrease the side effects are one the goals of may cardiac and industrial researches. To study the electric fieldinduced transmembrane potential distributions in cardiac tissues in response to the applied pacing and defibrillating currents, biophysical modeling studies have been performed using either idealized one-dimensional models of cardiac tissue or three dimensional periodic myocardia

159

(Weidmann, 1970; Robert and Scher, 1982; Plonsey and Barr, 1985; Blanchard et al.,1989; Krassowska et al., 1990,). The numerical results from the modeling yield the analytical relation between the large-scale field parameters and the small-scale membrane potentials. There is also a large body of effort focused on measuring cellular membrane potential in response to external applied fields. Dillon (1991) developed a novel technique using a voltage-sensitive dye to transduce the cardiac membrane potential into fluorescent emission and employing fiber optics to acquire these optical signals, thus permitting action potentials to be recorded without interference by the large interstitial electrical field produced by defibrillation shock (Dillon, 1991; 1992). A floating glass microelectrode was used to determine the effects of defibrillation shock on the action potential duration in the in vivo beating heart (Zhou et al., 1991; Knesley, 1992) and to evaluate the importance of the interstitial potential to the transmembrane potential measurement in ventricular muscles (Knisley et al., 1991).

The extracellular electric field-induced membrane potential changes are well described by Poisson's equation of electrostatics. With some modifications, the electric field-induced redistribution of the charged cell surface components has been shown to produce major alterations in membrane potentials over a time course of minutes to hours (Gross et al.,1986; Ehrenberg et al., 1987; Gross, 1988). However, all direct orindirect measurements ofthe electric field-induced changes in the membrane potential to date have been performed in a time range much shorter than the surface charge redistribution time (Benz and Zimmerman, 1980; Gross et aI., 1986; Gross, 1988;) thus it is not surprising that surface charge redistribution effects have not yet been observed experimentally.

Moreover, after clinical and experimental transthoracic defibrillation, undesirable collateral effects such as arrhythmias, immediate refibrillation, and other signs of dysfunction can exist. These electric field-induced dysfunctions have been demonstrated in the heart cells of several species. One example is the use of cultured chick embryo myocardial cells exposed to a high voltage electrical field stimulation (Jones and Jones, 1982; Jones et al., 1987) Enzymatically isolated frog cardiac cells have also been used with a patch clamp to study electroporation on cell membranes (O'neil and Tung, 1991; Tungetal., 1992). The current hypothesis is that the shockinduced, prolonged depolarization of the cell membrane is caused by the formation of transient sarcolemmal microlesions in ventricular cells. These microlesions take the form of an increased permeability of the cell membrane and allow a large, non-specific leakage and ionic exchange among the interior and the exterior of the cell (Jones and Jones, 1982; Jones et al., 1987).

Recently, the effects of a high-voltage electrical field on cell membranes of skeletal muscle fibers have drawn more and more attention in the clinical diagnosis and treatment of electrical trauma patients (Lee and Kolodncy, 1987). Beside the huge transmembrane currents induced thermal damages, membrane e1ectroporation has been considered as one of the mechanisms involved in electric injury. Current experimental results have confirmed that electroporation of cell membranes results in leakage of K+ ions and cardiac enzymes out of the cells (Miles and Zipes, 1991), osmotically driving water into the cell (Tsong, 1991), overloading intracellular Na + and Ca2+ ions (DeMello, 1982; Patton et al., 1984), and finally leading to cell necrosis and death.

In addition to the studies of membrane electroporation, mainly the phospholipid bilayerbased conformational changes (Chang and Reese, 1990; Tsong, 1991), there is a growing interest in understanding how the high-voltage electric field affect the structures and functions of the membrane proteins (Chen and Lee, 1994, b, c; Abrmov, 1996). So far, the electric field-induced conformational changes in the membrane proteins have not been thoroughly studied. In other words, cell viability has been studied in much greater detail than cell functions. Membrane

160

phospholipid bilayer has been focused on, and little attention has been paid to the membrane proteins.

2. Electro-Medicated Leakages in the Na+/IC ATPases

Pioneer work on the electrical shock-induced conformational changes in membrane proteins was done by Tsong and Teissie on the NaIK ATPase using erythrocyte cells (Teissie and Tsong, 1980). After being shocked by an intensive electrical field in a time range of microsecond the red cell membrane was perforation (Kinosita and Tsong, 1977). They found that in a low inoic medium at least 35% of the pores was related to the leakage through the NaIK ATPase in the cell membrane. That is because the shock-induced increase in the membrane conductance could be partially blocked by a specific inhibitor, ouabain, or by a specific cross-linking reagent, Cuphenanthroline, of the ATPase. They attributed this ionic leakage to the electroporation on the NaIK ATPase in cell membranes. In other words, the NaIK ATPase protein is one of the sites of the pore formation on the cell membrane. Recently, Abramov et al. (I996) performed in vivo studies on changes in the functions of rat sensory nerves and skeletal muscles after an intensive electrical shock. They found that 4 ms shock pulses of 500 V/cm can significantly reduce the magnitude of the action potential and increase the latency period of the action potential. They predicted that these alterations in nerve functions may be due to some conformational damages in the membrane proteins.

3. Methods to Study the Field-Induced Effects on Cell Membranes

1) Traditional Methods

Kinetics of membrane electroporation involve the initial electrical breakdown of the cell membrane and the secondary effect, cell lysis, processes that occur from within nanoseconds to subseconds (Tsong, 1991). Any structural changes in membrane proteins involved in electrical injury are resulted from the extraordinary force on proteins' charged particles exerted by the high intensity electric field. Those field-induced conformational changes in either the membrane phospholipid bilayer or the membrane proteins must occur during the electric shock. Therefore, the techniques used for monitoring cellular responses to electrical shock must be fast as well as accurate, both during and after the electric shock.

Traditional experimental studies of electrical damages in the skeletal muscle fibers involve morphological investigations using an electron microscope with freeze fracture (Chang and Reese, 1990) and permeability studies employing fluorescent dye (Lee et al., 1992; Chen et al., 1992). Radioactive staining methods has also been employed to investigate membrane electroporation and the field-induced effects on the Na+/I(+ ATPase (Teissie and Tsong, 1980; Serpersu and Tsong, 1983). All these methods involve measurements taken after the electrical shock has occurred and have time resolution on the order of 10-1 second or less.

An alternative method permitting higher time resolution uses a fluorescent, voltagesensitive dye to monitor membrane potential. Kinosita and co-workers (Masahiro et al., 1991) have described a sub-microsecond optical measurement capability in studies of electroporation in sea urchin eggs. They found that the membrane conductance gradually increased on the microsecond time scale. One of the important results was that the permeability of electroporated eggs and liposomes was markedly asymmetric. A common disadvantage of these approaches is

161

that changes in the cell membrane cannot be monitored during the electric shock, especially, at the onset ofthe electric field. Consequently, the field-induced increase in membrane conductance cannot be resolved into its components, corresponding to the membrane phospholipid bilayer and to the membrane proteins, respectively.

Direct measurements ofboth membrane potential and transmembrane current require more refined techniques such as "charge injection" or "charge relaxation" and voltage clamp. The charge injection/relaxation method pennits a time-resolution for transmembrane current in a nanosecond range (Benz and Zimmermann, 1980, 1981; Chernomordik et a/., 1983). A cell membrane is charged with high-voltage current and then discharged. The rate of discharging curve depends on the membrane's capacitance and conductance. The response time is fast, but the injection/relaxation method can only monitor the cell transmembrane current immediately after, not during, the electrical shock. There is another disadvantage of the injection/relaxation method. During the recording of the transmembrane current, the membrane potential decays exponentially, both the transmembrane current and the membrane potential vary during the measurement, making it difficult to study the current-voltage relationship.

Voltage-clamp techniques involve clamping a constant voltage across the cell membrane and simultaneously measuring the transmembrane current. The capability of monitoring changes in transmembrane current during the electrical shock makes it possible to study the dynamic changes in membrane conductance. Moreover, different components in the transmembrane currents, such as resistive current, capacitive current, linear current, and non-linear current, each corresponding to the function of a specific membrane protein, have different time course and voltage dependence. Therefore, the transmembrane current monitored during the shock using a voltage-clamp technique can be resolved into different components according to their kinetics and dynamics.

The voltage clamp method has been used to study electroporation in artificial lipid bilayer preparations (Abidor et a/., 1979; Chernomordik et a/., 1987; Glaser et al., 1988), but it is not generally used with live cells. One difficulty is that the high internal resistance of the

1\.1ACR()'PATCH

I T ICell

Figure 5. Configurations of macro-patch and micro-patch voltage clamps, and three pathways for electrical current flow. Taken from O'Neil et al., 1991.

162

microelectrodes limits the amount of injected current because of the large voltage droponthe

electrode. Recently, Tung and his colleague used a cell-attached patch clamp method to measure

changes in membrane patch conductance in response to DC voltage pulses (O'neil and Tung, 1991; Tovar and Tung, 1992). Single viable ventricular cells were obtained from whole frog heart via enzymatic dissociation. The investigation of the e1ectropermeabilization of heart cell membrane was conducted using two different cell-attached methods: macro-patch and micropatch. The study was launched using the macro-patch method which was modeled after the experiments performed by Zimmerman et al. (1980b) in which cells were e1ectropermeabilized as they passed through a modified Coulter cell counter device. Figure 5 shows the configurations of the macro-patch (whole cell patch clamp) and micro-patch, as well as electrical current pathway. There are three pathways for electrical current flows: capacitive current passing through the wall of the glass pipette; the resistive current through pathway between the cell patch and pipette tip; and current passing through the cell membrane.

The advantage of using micropipette electrode is that small cells, such as cardiac cells, can be studied, which have a length of only a few hundred micro-meter. However, when study the skeletal muscle fibers, it may not suit because of the long skeletal muscle fibers with a length up to centimeters. In addition, an intensive electric field-induced transmembrane current is much (a few orders) larger than the physiological membrane current. This large current passing through the large resistance inside the microelectrode (generally in an order of megohm) may cause a nonnegligible voltage drop on the electrode and result in a long rising phase in building up the membrane potential.

2) Improved Double Vaseline-Gap Voltage Clamp Technique

For skeletal muscle fibers, an elegant method, double Vaseline-gap voltage clamp was developed by Hille and Campbell (1976). Today, this method have been widely used to study ionic channel currents and intramembrane charge movement currents in the skeletal muscle fibers (Kovacs et ai., 1983; Irving et ai., 1987; Hui and Chen, 1992). The basic concept of this technique is to inject current into the cell at one end pool and monitor changes in the membrane potential at the other endpool.

When study the high-voltage electric field-induced damages in the skeletal muscle fibers, this technique emerges some disadvantages. Because of the cells interior resistance, the voltage drop on these intracellular resistance results in a non-uniform membrane potential distribution in the central pool. This voltage drop becomes non-negligible in the study of electrical injury on cell membrane. The ratio of the highest membrane potential over the lowest membrane potential in the central pool has been estimated as 12.5, shown in the lower right panel of Figure 6. In other words, if the membrane potential at the end pool 1 is clamped at 300 m V, the membrane potential in the end pool 2 may be more than 3 V. In addition, the membrane capacitance current, corresponding to the rising phase of the electric field is transiently much larger than the steadstate transmembrane current. This transient capacitive current will cause a transient voltage-spike at the current injecting endpool, which makes it difficult to determine the current-voltage relationship, especially the voltage-threshold for damaging cell membranes.

Recently, we modified the configuration of the traditional method and made the improved

double Vaseline-gap voltage clamp technique more suitable for the study of electrical injury in the

163

skeletal muscle fibers (Chen and Lee, 1994, a). The improved voltage-clamp configuration with the double Vaseline gap chamber is shown in the upper panel of Figure 6.

Of ]0 vi ........•................

.................................. .2 ! I I I

v v vi .2

Figure 6. Configuration of the improved double Vaseline-gap voltage clamp and schematic plot of the membrane potential distributions. Upper panel is the configuration. Lower left and lower right panels are

the membrane potential distribution along the fiber axis for the improved and the traditional voltage clamp, respectively. Taken from Chen el 01., 1994, d.

In contrast to the traditional configuration, the pools of current injection and potential measurement are electrically connected to eliminate the transient overshock on the cell membrane, and to even out the distribution of the membrane potentia! in the central pool. To overcome the voltage drop induced by the cell's interior resistance, we have developed a series resistance compensation circuit to compensate the voltage-drops on the fiber segments underneath the Vaseline partitions.

This improved double Vaseline-gap voltage-clamp technique fully eliminates the transient over-shock at the rising phase of the electric field. Using our new clamp configuration and the modified chamber (the widths of the partitions and central pool reduced to 100 ~m and 300 ~m, respectively), the ratio of the highestllowest membrane potentials in the central pool is only 1.8, seven times smaller than that obtained with the traditional configuration. The schematic distribution of the membrane potential along the cell axis is shown in the lower left panel of Figure 6.

Using this improved double Vaseline-gap voltage-clamp technique, we can keep the resistance of the voltage-applying-electric circuit within less than one kilo-ohms, that is several orders in magnitude lower than that of a microelectrodes voltage clamp. Therefore, this method allows us to deliver a strong, pulsed electric field to cell membrane using a voltage-clamp, and

164

to simultaneously monitor changes in the transmembrane current during the electric shock, especially at the rising phase of the shock pulse.·

4. Direct Observation of Membrane Electroporation Current

1) Membrane Electroporation Current

Before the discussions of the effects of a supramembrane potential on the membrane proteins, it is necessary to understand how the electric field affect the membrane phospholipid bilayer. Recently, we have directly measured the electric field-induced dynamic changes in the membrane conductance during the shock pulse. Figure 7 shows a 4 ms, 500 m V shock pulseevoked transmembrane current on a skeletal muscle fiber. The reason of choosing 4 ms duration of the shock pulse is to mimic an exposure to a power-line-frequency electric field. A 4-ms duration pulse and a half-cycle sinusoid current at commercial (60 Hz) power frequency with the same RMS value have the same energy.

nA

o

-5128 5

I 4ms

,'-------' nA

o

lOms -5128

-90mV

-590mV

5 10 ms

Figure 7. Transient transmembrane current recorded during a high electric shock. Top: a shock pulse of 4 ms, -500 m V, and lower left: recorded response transmembrane current. After computationally removing the capacitance and leakage currents of the cell membrane, the pulse-induced membrane electroporation current is shown in the lower right Taken from Chen et 01., 1994, a.

The top of Figure 7 shows the shock pulse. The trace in the lower left panel is the evoked total transmembrane currents recorded simultaneously during the shock. The transient jump with an exponential decay is capacitance current. The transmembrane current then rises integrally throughout the pulse. To identify the pulse-induced leakage currents, the capacitance and normal leakage currents can be differentiated by using the PIN method. The pulse-induced electroporated current is shown in the lower right panel. Compared with morphology studies of the shocked cell

165

membrane using an electron microscope (Chang, 1990),the pulse-induced membrane leakage current is related to a formation of electro-mediated pores or pore-like structure in the cell membrane.

From the lower right panel, the electro-mediated current keeps monotonously increasing until the end of the shock pulse. Starting from the falling edge of the shock pulse,thepulseinduced leakage current decreases significantly within milliseconds and reaches a small and relatively stable plateau. This recovery time-course is consistent with the results from other investigators (Tsong, 1991). Considering the recovery time course (in an order of minutes) obtained from measuring the restoration of the post-shock membrane holding current, it appears that there are at least two kinds of electro pores with the recovery time courses of milliseconds and minutes, respectively. Interestingly, the short-lived pores contribute the most to the changes in membrane conductance while the secondary pores contribute relatively little.

2) Potential Threshold for Membrane Electroporation

Our experimental results showed that the noticeable membrane leakage current of skeletal muscle fibers was occurred when the cell membrane was shocked by a 4 ms pulse to a membrane potential from -250 to -300 mY.

For cardiac muscle fiber, Tung et aI., analyzed the distribution of voltage levels sufficient to cause electroporation in enzymatically isolated frog cardiac cells, using the cell-attached patchclamp techniques with rectangular pulses similar to these used in experimental studies of cardiac defibrillation (Tung, 1992). Five-millisecond monophasic or ten-millisecond biphasic symmetric and asymmetric rectangular pulses steps from 200 to 800 m V. The membrane conductance was continuously monitored by a low-voltage pulse train. They found that over 20% of their experimental cells showed increase membrane conductance after a single 5 ms, 300 mV shock. They concluded that 5 ms monophasic or 10 ms biphasic pulses can permeabilize the cardiac cell membrane at transmembrane potential of300 m V, which is consistent to our results from skeletal muscle fibers. The membrane electro po ration threshold for cardiac muscle fibers is about 300 m V.

3) Depolarization of Membrane Resting Potential

To study high intensity electric field shock-induced depolarization of the cell membrane resting potential, we designed and performed current-clamp experiments to monitor transient changes in membrane potential during and after the electrical shock. A series of large-current, rectangular pulses was delivered to shock the cell membrane, and any transient changes in membrane potential during and following the pulsed shock were simultaneously monitored. The membrane resting potential in these studies shifted in the depolarization direction by about 7±4 mV following a single 4-ms, 1 JlA shock pulse, which is a typical transmembrane current during a -600 m V pulsed shock in a voltage-clamp mode.

5. Functional Reductions in Ion Channels

1) Conductance Reductions in the K+ and Na+ Channel

A series experiments have been performed to study the effects of supramembrane potential on ion channel proteins in skeletal muscle fibers (Chen and Lee, 1994, b, 1998, a, c, d). Before

166

and right after the cell membrane was exposed to a 4 ms, 500 m V supramembrane shock pulse, a 120 mV stimulation pulse was applied to the membrane. The pre- and post-shock voltage-gated Na+- and the delayed rectifier K+-channel currents are shown in the left upper and left lower panels of Figure 8, respectively. Both the peak inward Na+ current and the steady-state outward K+ current were reduced by a single shock pulse. Compared with the upper panel, about 20% of the Na+ channel currents and 30% of the delayed rectifierK+ channel currents were eliminated

by a single 4-ms shock pulse of -500 mY, as shown in the low panel.

1024

nA

0 ZSS

I I InS 5 25 50

nA 1024

0

I InS 5

nA 2S 50

Figure 8. Effects ofa 4 ms, -500 mVpulse on the voltage-gated Na+- and K+-charmel currents. The left upper and left lower panels shows the resolved channel currents responding to a 30 ms, 120 m V stimulation pulse recorded before and after electric shock, respectively. The bath solution was then changed to with 1TX and TEA to block the channels and the recorded residual channel currents are shown in the right

panel. It indicates that primary current in the left panels are the Na + - and K+ -channel currents. Taken from Chen et al., 1994, h.

Next question is how a large imposed electrical shock affects the binding sites in the channel proteins for neurotoxins tetraethylammonium ion (TEA) and tetrodotoxin (TTX). What fraction of the current recorded after the electric shock (shown in the left lower panel) can be attributed to the Na+ and K+ channel currents? To answer these questions, we changed the bath solution in the central pool to the solution with 50 mM TEA and 1 JiM TTX immediately after the post-shock measurements. The same stimulation pulse was again applied to the membranes, and the evoked channel currents were found to be negligible (right panel). This result indicates that following a single 4-ms pulse of -500 mY, the reduced, post-shock voltage-gated Na+ and K+ channel currents can still be substantially blocked by TTX and TEA, respectively. In other

167

words, the neurotoxin, TTX and TEA, binding effects on the Na+ and K+ channels are not affected by the electrical shock.

2) Shock Effects on the IC Channel Conductance

To study the more details of how an intensive electric field affect the delayed rectifier K+ channels in cell membranes, a stimulation pulse sequence was employed which consisted of30 consecutive pulses holding the membrane potentials in a range from -65 m V to 0 m V. The evoked delayed rectifier K+ channel currents were resolved and are shown in the left panel of Figure 9, where theNa+ channels were blocked by TTX.

After shocked by a 4-ms pulse with a supramembranepotential of -500 mY, the same stimulation sequence was applied again to the cell membrane and the recorded K+ channel currents are shown in the right panel of Figure 9. By comparing the pre- and post-shocked channel currents, one can easily notice that the K+ channel currents were reduced by a single brief electric shock. For this fiber, the maximum value of the pre-shock K+ channel currents, about 220 nA, was reduced to around 175 nA, an approximately 20% reduction.

Figure 9. Shock field induced changes in the delayed rectifier K+ channel currents. Right and right panels show the channel currents responding to the same stimulation pulse sequence (25 ms; holding the membrane from -65 to 0 m V) recorded before and after shocked by a 4 ms, -500 m V pulse, respectively. Taken from Chen et 01., 1998, c.

The K+ channel currents in the left and right panels are plotted as functions of the membrane potential shown in Figure 10, represented by dark circles and open squares, respectively. The slope of the post-shock I-V curve is lower than that of the pre-shock curve, indicating the shock pulse-induced reductions in the K+ channel conductance. For this fiber, the pre- and post-shock K+ channel conductance is 4.4 IlS and 3.5 IlS, respectively. Approximately 200.10 of the channel conductance was reduced by a single 4 ms, -500 mV pulsed shock.

3) Potential Thresholds for Membrane Electroporation and for Damaging Channel Proteins

The electroporation current studied in this chapter is a pulse-induced, non-specific membrane leakage current (Tsong, 1991). Experimental results from our and other laboratories

168

250

• 200 • •• •

• rP ! 150 ••• r:;i1P

'a •• r:PC

~ .·CC d 100 .~pc 1 ~c co 0° 'fi 50 ~ ~r;jl

C. C~·

0 coacOQ.

·50 ·70 ..so ·50 -40 -30 ·2D ·1D 0

Membrane Potential (mV)

Figure 10. The K+ channel current asa function of the membrane potential. Dark cycles and open squares represent the channel currents recorded before and after the electric shock, respectively. Taken from Chen et 01., 1998, c.

showed that the field-induced leakage current could be measured when the cell membrane is shocked by a 4 ms supramembrane potential pulse in a range between -250 mV and -300 mV (Tung el al., 1992; Chen and Lee, 1994, b).

The results also showed little difference of the voltage-gated Na+ or K+ channel currents before and after shocked by a single, 4-ms pulse up to -400 m V, which exceeds the electroporation threshold of the cell membranes. Thus, the membrane potential threshold for damaging the voltage-gated channel proteins is higher than that for damaging phospholipid bilayer. The reason for this discrepancy is unknown. One explanation may relate to the structure of the phospholipid bilayer as a supramolecular assembly held only by forces of hydration without chemical bonding, whereas, the amino acids and subgroups in the channel proteins are assembled with strong chemical bonds. Differences in charge densities and distributions may also account for the different electro-mechanical forces induced by the external electric field.

It is necessary to notice that in this comparison of membrane potential thresholds for ion channel damage and for membrane electroporation, the measured electroporation current is only attributed to the category of fast recovery pores. In fact, the membrane potential threshold for damaging the K+ channel proteins is comparable to that for generating permanent electropores in the cell membrane.

6. Thennal Damages Versus Electroconformational Damages in Membrane Proteins One of the possible mechanisms involved in electrical injury is Joule heating. The direct

effects ofJoule heating in damaging cells and tissues have long been recognized as denaturation

169

of the cell membrane and membrane proteins. The characteristic of Joule heating-induced damages in cell membrane is that the degree of damage is proportional to electrical energy consumed in the cell membrane, which is generally directly related to the field-induced transmembrane currents. (Rouge and Dimick, 1978; Bingham, 1986; Lee et af., 1988).

In addition to Joule heating damages, electroporation in cell membranes was also considered as one of the mechanisms of electrical trauma (Lee and Kolodney, 1987; Jones and Jones, 1987; O'Neil and Tung 1991; and Chen et af., 1992). In contrast to Joule heating, fieldinduced supramembrane potential, instead of the huge transmembrane current, plays a significant role in membrane electroporation (Kinosita, 1977). As mentioned previously, membrane electro po ration has been postulated mainly as electro conformational damages in the phospholipid bilayer induced by supramembrane potential (Tsong, 1991).

Now, a question is raised whether the high-intensive electrical field-induced functional reductions in the membrane proteins, such as the K+ channels, are resulted from electroconformational damages or from field-induced Joule heating. The structural damage caused by a supramembrane potential may involve movements of the intramembrane charged particles, or reorientations ofthe equivalent dipole moments in the channel proteins; while the Joule heating caused by a huge transmembrane current may involve a local temperature increase resulting in an electrical bum in the narrowest pores of the protein channels.

It is necessary to identity among the supramembrane potential-induced electroconformational changes and the huge transmembrane current-mediated thermal effects, which one is the dominant mechanism involved in the membrane proteins damage in electrical injury.

1) Distinguishing Electroconformational Changes and Thermal Damages in the K+ Channel Proteins

Because of difficulty in measurements oflocal, transient changes in the temperature ofthe electrical injured cell membranes, we estimated the electrical energy consumed in the cell membrane during the electrical shock. According to Joule's Law, the electric field-induced evolution rate of heat on the cell membrane should be equal to the shock pulse-induced electric energy consumed on the cell membrane. We can estimate the consumed energy by integrating the transmembrane currents with time during the shock pulse and then multiplied by the applied membrane potential difference. Since most of the external field-induced potential difference is across the cell membrane, the estimated electric energy should be mainly consumed in the membranes, representing the transmembrane Joule heating.

A stimulation pulse sequence consisting of 15 consecutive pulses holding the cell membrane potential from -60 to 10m V with 5 m V increments was used to measure the K+ channel currents before and after the electric shocks. For this fiber, the control K+ channel currents (not shown) taken before any shock exhibited a maximum value of 168 nA. The fiber was then shocked by a +350 mV pulse and the evoked transmembrane currents are shown in the left panel of Figure 11. Right after the shock, the K+ channel currents were recorded, as shown in the right panel of the Figure. Allowing the fiber relaxed for at least 10 minutes as evidences for full recovery that both the holding current and the K+ channel currents were fully restored to the control value. Finally, the fiber was shocked by a -350 m V pulse. The evoked transmembrane currents and thereafter recorded K+ channel currents are shown in the left and right panels of Figure 12, respectively.

170

Sl'12tOG1 200

4000

~ j 0

-4000

0 Tunc (ms) S o Timc(ms) 20 40

Figure 11. The totaI transmembrane currents induced by a 4 ms, +350 m V shock pulse Qeft) and the postshocked K,+ channel currents responding to a stimulation pulse sequence (right). The shock pulse-induced peak value of the transmembrane current is about 2200 nA, while the post-shock channel current show little changes compared to those recorded before the shock (not shown). Taken from Chen et al., 1998, a.

Because of under a voltage clamp, the membrane potentials during the two shock pulses were held at the sarne magnitude. However, the responding transmembrane currents are significantly different due to the opposite shock polarity. In Figure II, the transient capacitance current is followed by a huge K+ channel current plateau, which is the characteristic of the delayed rectifier K+ channel currents. At end of the shock period, the transmembrane currents was slightly increased which may be due to membrane electroporation. The peak value of the transmembrane current for this fiber (except the transient capacitance currents) is about 2200 nA (left panel of Figure 11).

200 Sl'lZt013

100

~

l'~~~~~b~ o S Timc(ms) o 20 40 Timc(ms)

Figure 12. The total transmembrane currents induced by a 4 ms, -350 m V shock pulse Qeft) and the postshocked K+ channel currents (right). The shock pulse-induced peak transmembrane current is only less than 200 nA, but the K+ channel currents were significantly reduced. Taken from Chen et al., 1998, a.

171

In the left panel of Figure 12, since the fiber being hyperpolarized, no channel currents are observed. The negative shock pulse only generated less than 200 nA leakage current (left panel of Figure 12). In contrast, the positive shock pulse which opened the K+ channel, generated a ten times more current across the membrane.

The maximum value of the post positive-shocked K+ channel currents, shown in the right panel Figure 11, is 165 nA, only a slightly decrease from the control. This result indicates that the positive shock pulse had little, if any, effects on the channel currents. However, the negative shock pulse significantly reduced the channel currents to 140 nA, shown in the right panel of Figure 12. We can integrate the resistive transmembrane current in order to calculate the electric energy consumed on the cell membrane. The +350 m V shock pulse drove almost 10 times more charges across the cell membrane than the -350 mV pulse, but cause only 1/9 reduction of the channel currents.

In fact, when we cross-compare the membrane potentials of the shock pulses, the evoked transmembrane charges, and the resultant reductions in the channel currents in response to the shock pulses of -350 mV and +500 mY, the results are even more conclusive. The Figure 13 shows the comparison of the shock pulse-induced Joule heating on the cell membranes and the percentage-reductions in the K+ channel currents in response to -350 mV and +500 mV shock pulses for thirteen fibers . When compared with -350 mV shock, a +500 mV shock pulse drove about 19 times transmembrane charges and thereby generated 27 times Joule heating on the cell membrane, but only caused less than one half of the reduction in the K+ channel current.

14 25 0 ""' 12 "$.

20 '-'

'0' c; ~ <1)

0 10 .... ... ..... ::J 0 § 0

15 v -=- 8 c:

11 c: CIS

.c: ::r:: 6

0

OJ 10 ~ :i 0 OJ ..... -5 ] 4 '-0

CIS 5 c: ... OJ .9 s:: 2 OJ 0 <.:I ::J

"0

0 0 <1)

P::!

+500 mV ·350 mV

Shock pulses (mY)

Figure 13. A comparison of the shock pulse-induced Joule heating effects with the induced percentagereductions in the K+ channel currents when shocked by 4 rns pulses of -350 m V and +500 m V, respectively. The dark columns represent the shock pulse-induced Joule heating effects in a scale ofnanojoules shown as the left ordinate. The open columns represent the percentage-reductions in the K+ channel currents shown as the right ordinate. The bars represent standard deviations. Taken from Chen et al., 1998, a.

172

These results clearly indicate that there is no direct correlation between the shock fieldinduced Joule heating effects on cell membrane and the reduction of the K+ channel currents. The damages in the K+ channel proteins should not be attributed to the field-induced huge transmembrane currents, therefor the current-mediated thermal effects on the cell membranes. In contrast, the protein damage level closely depends on the supramembrane potential, the potential magnitude and polarity. In other words, the proteins damages may mainly due to the supramembrane potential-induced electroconformational changes in the proteins structures.

2) Ie Channel Functional Reduction Is Not Directly Related to the Local Joule Heating

There could be two types of Joule heating. The first type means heating of whole membrane in general, which is directly related to the transmembrane currents. The second type is heating of specific membrane proteins, in this case the K+ channels. As discussed above, our results clearly exclude the possibility of the first type Joule heating as the cause of reduction in K+ channel activity. For the second Joule heating on the K+ channel proteins, we need to fully understand what our integrated ion flux means.

Even though we can not quantitatively identify the pure current passing through the K+

channels during the shock pulses, it does not affect our qualitatively analysis. As an example, let's go back to the comparison of the channel function reduction caused by a +350 m V and a -350 mV shock pulses. Since the same magnitude of the shock pulses, the linear leakage current are the same, which by name is proportional to the membrane potential. We now only have to distinguish the channel currents and the membrane electroporation currents.

From the left panel of Figure 11, it is clearly shown that the majority of the resistive current responding to a +350 mV pulse was carried by the K+ channels. We identified and estimated according the kinetics difference that more than 95% of the resistive currents are attributed to the channel currents. For the -350 mV shock pulse, we do not know what percentage of the resistive current passing through the K+ channels. For the worst situation, let us assume all of the resistive currents passing through the channels (even though it is impossible), we still have the similar result. In other worlds, a +350 mV shock pulse evoked about 9 times current passing though the K+ channels, therefor generated 9 times local Joule heating on these proteins when compared to the -350 mV shock pulse, but only caused 1/9 of the reduction of the channel activity.

These studies show that supramembrane potential shock pulse-induced reductions in the K+ channel functions are not directly related to neither the globe transmembrane currents nor the channel currents. Therefore neither the cell membrane Joule heating nor the Joule heating on the K+ channel proteins directly causes the protein damages. In other words, these experimental results provide strong evidences that during an exposure to a high intensive electrical field, the field-induced supramembrane potential (magnitude and polarity) plays a dominant role in damaging the K+ channel proteins. The thermal effects caused by the field-induced huge transmembrane current and channel currents, therefore the Joule heating effects, play a secondary, trivial role.

It is necessary to point out that the supramembrane potential we studied is up to 500 m V. We can not rule out the possibility that when the intensity of the field further increases, the thermal effects may playa more and more important role in damaging the membrane proteins.

173

7. Electroconformational Changes in Membrane Proteins

So far, very few detailed mechanisms involved in protein damages in electric injury have been reported. The possible mechanism underlying the field-induced structural damages in the delayed rectifier K+ channel proteins in cell membranes are not clear. Some interesting questions arose, whether we can experimentally provide evidence supporting the proteins' electroconformational damage theory, and more specifically, estimate changes in the K+ channel gating system.

Since the protein damages can not be simply attributed to the Joule heating effect, and the damage level is directly related to the magnitude of the supramembrane potential, a possible explanation is that the huge intramembrane field-strength or the supraphysiological membrane potential difference may exert tremendous electrical force against the charged particles in the membrane proteins. Normally, these charged particles function as the voltage-sensors in the proteins. There are two kinds of voltage-sensors. One is for muscle fiber excitation-contraction process, i.e. dihydropuridine receptors in the transverse (T) tubular membranes, which are sensitive to the membrane action potential and somehow trigger the Ca2+ release from the sarcoplasmic reticulum (SR). The other voltage-sensor is for each individual kind of membrane proteins, such as the gating system in the r channel proteins. The field-induced supramembrane potential will apply a huge electrical force on these voltage-sensors resulting in dysfunctions of the membrane proteins.

To obtain experimental evidence, there are two kinds of studies needed to be done. One is to understand the effects ofa supramembrane potential on the membrane proteins' movable charge particles by directly measuring the non-linear, intramembrane displacement currents. More specifically, quantitative studies of the changes in the channel gating system will give us more detailed insight into the channel proteins' conformational damages. These studies can be done by estimating the limiting numbers of the gating charged particles in the K+ channels before and after an electric field shock.

1) Reductions in the Intramembrane Charge Movement Currents

Reductions in the Charge Movement Currents. The intramembrane charge movement current was first measured by Chandler and Schneider in 1972 (Schneider and Chandler, 1973) using the three microelectrode technique. This current was now defined as the fast component, QII' of the change movement currents. In 1977, Adrian and Peres (1977) reported another component, Qy, of the charge movement current, the slow hump component. Since then, charge movement has been widely investigated regarding to its structures and functions in excitationcontraction coupling process (Huang, 1982; Rios and Brum, 1987; Chandler and Hui, 1990; Chen and Hui, 1992; Pape eta/., 1996; Hui and Chen, 1997).

We recently investigated the effects of a pulse with a supraphysiological membrane potential on the charge movement currents (Chen et a/., 1998,d). The upper panel of Figure 14 shows the measured charge movement currents responding to a series of stimulation pulses holding the membrane potential from -60 to 0 mY. The transient peak with an exponential decay is the fast component of the charge movement current, QII' while the slow component, Qy, is marked as a "hump" following the fast component. The dot line shows the position of the peak

174

for each fast component, QII. The slow hump component, Qy, started to be noticeable at the second or the third traces.

~ .......

~ =' u

'6 ~ ~

j ~70mV

U Time (ms)

0 25 50 75

~ .... ~ ----------.1. ___ =' u

I ~

j .70mV

U Time(ms)

0 25 50 75

Figure 14. Supramembrane potential shock-reduced charge movement currents. The stimulation pulse sequence consists of nine 40 ms pulses holding the membrane potentials from -70 to 10m V. The upper and lower panels represent the change movement currents recorded before and after shocked by a 4 ms, -350 mV pulse, respectively. Taken from Chen et al., 1999, b.

The charge movement currents elicited by the rising phase of the stimulation pulse is a "on" charge movement, while those responding to the falling phase of the stimulation pulse is a "off' charge movement. According to the definition ofthe intramembrane charge movement, the amount of the "on" charge movements should be equal to the amount of the "oft" charge movements. Both the "on" and "off' charge movements can be obtained by integrations of the "on" and "off' charge movement currents with time, respectively.

The lower panel of Figure 14 is the charge movement currents recorded after the fiber was shocked by a 4 ms, -350 mV pulse. The peak value of the slow component, Qy is significantly reduced by the shock. The decrease in the peak of the hump component indicates that the moveable charged particles corresponding to Qy .is reduced after the electrical shock. This can be shown as a down-shift of the plateau ofthe post-shock Q-V curves in Figures 15. In contrast, the fast component, QII' in the post-shock current traces of Figure 14 shows very little changes.

175

25 ~ 20 • • •

• ~ 15 • [J

[J [J c [J :g. [J

I 10 • [J

~ 5

• 0 C [J 0

0

-80 -80 -40 -20 0 20

Membrane potential (my)

Figure 15. Steady-state I-V CUlVes of the "on" charge movements: control (dark cycles) and shocked by a 4 ms pulse of -350 mV (open squares). The inset shows an integration of the "on" charge movement currents. Taken from Chen et al., 1999, b,

When cells were shocked by a -500 mV pulse, both the fast and slow components decreased, therefore the total charge movement, or the movable charge particles, were significantly reduced. This result clearly shows that the reduction in the charge movements depends on the magnitude of the supramembrane potential difference or the intramembrane fieldstrength. These studies lend us further credence to our hypothesis that supramembrane potential may affect the protein structures by altering the movability of the charge particles in the membrane proteins.

Time Delay in the Peak ofQT Component. Interestingly, in addition to reduction in the magnitude OfQT' the kinetics of the hump component of the charge movement current was also changed due to the electric shock. The shape of the post-shock hump component became broader· and the appearance of the hump became more delayed. From Figure 14, it is clearly to see that the time delay becomes longer in the post-shock charge movement currents than those in the preshock currents.

The time delay can be quantitatively described as the delayed-time which is defined as the time interval between the rising phase of the stimulation pulse and the peak of the slow component, QT' Both the pre- and post-shock time delays ofthe slow hump component, Q" are membrane potential dependent. The larger the membrane potential, the lesser the time delay is. At certain membrane potential the slow component, QT' will catch up the fast component, QII' The delayed-times in the hump peak of the slow component, Q, are plotted in Figure 16, as a function of the membrane potentials.

176

16 ~IimCdclay 0

14 ~ ! li 12

i • E 10 co 0 u

Oi 8 ~ • 0 8. 6 ... • 0 co ,..

• 0 .!! 4 u 0 ... • 0 u E 2 • t= •

0 -60 -50 -40 ·30 ·20 ·10 0 10 20

Memb!3lle potential (mV)

Figure 16. Time delay of the peak of the slow charge movement current component after a single electric shock. The inset shows the definition of the delay-time. The dark cycles and open squares represent the control and those after a 4 ms, -350 mV shock, respectively. Taken from Chen et al., 1999, b.

After electrical shock, the time delay became longer. Even in response to the last simulation pulse of + 10m V, the hump component is still slower than those in the pre-shock charge movement currents, with about 1 ms more delay. These shock pulse-induced kinetics changes imply that after shocked by a supramembrane potential-pulse, the moveable charge particles are more reluctant than those before the shock, if they are still moveable.

A Comparison of Membrane Potential Thresholds. Several laboratories have reported that membrane transient electroporation occurred when the membrane potential is held at -250 to -300 m V for 4 ms in frog skeletal muscle fibers (Chen and Lee, 1994, a) and in cardiac muscle fibers (Tover and Tung, 1992). It has also been shown that under a voltage clamp, the Na+ and K" channel currents decreased when the cell membrane was shocked by a 4 ms pulse of membrane potential of -400 m V.

Now, our results showed that a 4 ms, -300 mV shock pulse did not affect the charge movement currents at all, even though it is high enough to eJectroporate the cell membranes. This result indicate that the membrane electroporation threshold, at least for the transient electropores, is lower than the threshold for reducing the charge movement currents.

The discrepance between the potential ranges of reducing the K+ channel functions and of decreasing the slow component, QT' of the charge movement currents may be resulted from the voltage measurements. In the experiments of studying the charge movement currents, because of maximally blocking all of the ion channels, the shock pulse-evoked residual channel currents are negligible so that voltage drop on the series resistance is small. When study the electric shockeffects on the channel currents, no channel blockers was used so that the voltage drops on the series resistances induced by the large transmembrane currents can not be neglected.

Moreover, our results showed further difference between the two components of the

177

charge movements current (Chen et al., 1998, d). When the fiber was shocked by a 4 ms pulse of -350 mY, the fast component, Q~. of the charge movement current seems having no or little decrease, while the decrease in the slow component, Qy, is much more noticeable. These results indicate that the membrane potential threshold for reducing the fast component, Qp, is higher than that for the slow component, Qy. The reason behind the discrepancy in the damaging-potentialthresholds for the two components remains unclear.

2) Changes in the Gating Systems of K+ Channel Proteins

K+ Channel Conductance Reduction and the Channel Open-Threshold Shift. The charge movement currents we measured were resulted from the whole cell membranes. The gating currents of the ion-channel proteins are involved in the charge movement currents. It is difficult to identify the charge movement current specifically attributed to the K+ channel proteins. However, it is possible to estimate changes in the number of gating charged particles in the K+ channels. This can be done by comparing the channel open probability according to the Boltzmann theory before and after an electric shock (Chen and Chen, 1998, c).

From Figure 9, we have noticed that the pre-shocked, maximum K+ channel current is larger than that of the post-shock channel currents. But this is not always the case. When the cell membrane was stimulated by small pulses holding the membrane potential slightly higher than the channel open-threshold, the same stimulation pulse evoked larger post-shock channel currents than the pre-shock currents. In other words, this result indicates that the post-shocked, K+ channel open-threshold was more negative than that of the pre-shock channels.

Reduction in the Limited Gating Charged Particles in K+ Channel. Almers (1978) and Hille (1992) estimated the limiting number of gating charge particles in the K+ channels. They employed a simple two state model and studied the channel open probability. According to Boltzmann theory, they concluded that about six and five elementary movable charge particles in the Na+ and K+ channels, respectively, functioning as voltage-sensor. In order to better understand the shock field-induced electroconformation changes in the channel gating system, it is necessary to identify wether the number of the gating charged particles in the K+ channels are affected by a high-voltage electric shock.

Let us first start from a simple two-state model of channels: open and closed. According to thermodynamics, the possibilities of channel opening can be described by the Boltzmann distribution of the movable charge particles in the gating system. There are two kinds of energy that dominate the distribution of the gating charge particles. One is the protein conformation energy difference between the closed state, Ee, and the open state, Eo. Another is electrical energy carried by the external stimulation pulses. The electrical energy is -zeV, where V is membrane potential of stimulation pulses, e is the elementary charge, and z is the limiting number of movable gating charge particles or dipole moments (Almers, 1978, Hille, 1992). The total change in energy from the closed state to the open state responding to the stimulation pulse is Eo-Ee-zeV. The probability of the open channels among the total number of channels, P, can be described by the Boltzmann equation,

1 p= ----------

1 + exp[(Eo - Ec - ze V) I KT] (4)

178

where K is the Boltzmann's constant and T is absolute temperature in Celsius. When the membrane potential is very negative, lower and closer to the channel open-threshold, the second term in the denominator is much larger than the first term. The equation can be simplified to

p= exp[(zeV + Ec- Eo)/ KT] (5)

By taking the natural logarithm on both sides of the equation, the logarithm of the probability of the open channels can be expressed as a linear function of the membrane potential V.

Ec Eo ze InP=---+-V

KT KT KT (6)

Using this equation, the channel open probability expressed as the relative channel conductance of the K+ channels can be plotted as a function of the membrane potentials in a semilogarithm coordinate. Figure 17 shows the relative channel conductance versus the membrane potential from nineteen fibers before and after the electric shock. The relative channel conductance is the measured channel conductance normalized to the maximum channel conductance, which can be determined by stimulating the membrane at a potential much higher than the channel-open threshold. Thus, the relative channel conductance expressed as a percentage of the maximum conductance represents the probability of the open channels among the total number of channels.

o

4~--~,-----,-----,-----,-----.-----,

-60 ·50 40 -30 ·20 ·10 o

Membrane Potential emV)

Figure 17. Channel open probability as a function of membrane potential. The abscissa is the membrane potential in a linear scale and the ordinate is the channel open probability expressed as a normalized channel conductance drawn in a logarithm scale. Dark circles and open squares represent the pre- and postshock channel open probabilities, respectively. The two straight lines were obtained by fitting the rising part data to Equation 6. Taken from Chen et 01., 1998,. b.

179

In Figure 17, the pre- and post-shock channel open probabilities was obtained by normalizing the channel conductances to the maximal conductance recorded at +30 mV before and after the electric shock, respectively. The slope of the curve at the rising part indicates the number oflimiting charged particles, z. The larger the number of the limiting charged particles, the steeper the rising part of the curve is. Figure 17 shows that the curve slope of the pre-shock channels is steeper than the slope of the post-shock channels. It indicates that the number of limiting charged particles was reduced by the electric shock.

We can fit the rising parts of the two curves in Figure 17 to Eq. 6 as two separate straight lines. The slopes of the best-fit straight line for the pre-shock channel is 4.1 m V per e-fold (2.72) increase of the channel open probability. The value ofKT/e is about 24 mV, so that z = 24/4.1 =5.8, which means for the pre-shock K+ channels the limiting number of charge particles is about 5.8. The slope of the best-fit line for the post-shock K+ channels is 5.3, indicating that the limiting number of charge particles in the channel gating system was reduced to 4.5 after electric shock. A single 4 ms, -500 m V electric shock eliminated over 20% ofthe limiting charge particles in the K+ channel proteins.

It is necessary to point out when using channel conductance to express the channel permeability therefore the channel open probability, it implicitly admit "a instantaneous currentvoltage relation." Indeed, this is not always the case. In general, this is true only when the ionic concentrations crossing the cell membrane are the same and when only one kind of channels existence. However, studies from the GHK current equation showed that for voltage-gated Na+ and K+ channels, these factors only play a trivial role in the channel conductance steepness (Goldman, 1943; Hille, 1992). The maximum steepness of the voltage dependence in the channel conductance is mainly dependent on the channel gating system. In other words, we can attribute the steepness of the K+ channel I-V curve to the steeply voltage-dependent opening and closing of the K+ channels.

Intrinsic Relationship between the Gating Charged Particles and the Channel Open-Threshold. The experimental observations include that 1) a lower slope of the channel open probability as a function of the membrane potentials, and 2) a negative shift of the postshock channel open-threshold. The first observation has been used above to predict a reduction of the limiting number of the charged particles in the channel gating system according to the Boltzmann theory. The second observation has not been used in discussing the channels' electroconformational changes. Are these two observations independent? Are they together reveal the same conformational changes in the channel proteins?

Because of using channel conductance to represent the channel permeability and the channel open-probability in the two-state model, an instantaneous current-voltage relationship has been assumed. As mentioned before, we can only attribute the maximum steepness of the K+ channel I-V curve, obtained at a very negative membrane potential, to the steep voltagedependence of opening and closing the K+ channels. However, it is not necessarily true for the whole potential range.

In fact, the electric energy is not only used to relocate the gating charged particles, but also has to free these charged particles first. We now assume a multiple-state model by which charged particles move from different closed states to the final closed state, where they are ready to move to the open state. This can be expressed by a simple first approaximation,

Z= zo+ nV (7)

180

where Zo is the free charged particles when the applied external membrane potential is zero and n is a coefficient constant. The number of free charge particles. z. is a linear function of the membrane potential. V. By substituting Equation (7) into Equation (4). the channel open probability can be expressed as follow:

p= 1 1+ exp{[Eo- Ec-(zo+ nV)eV]/ KT}

(8)

Then. we fitted the data of the pre- and post-shock channel open probabilities shown in Figure 17 to Equation (8) (Chen and Chen. 1998. c). The two best-fit curves are shown in Figure 18. The ratio of the number of gating charged particles. z. at a membrane potential of -50 mVis 4.97/6.53. where the nominator and denominator represent the post-and pre-shock K+ channels. respectively. The ratio value is about 76%. indicating 24% of the channel gating charged particles was eliminated by the electric shock. which is relatively consistent with the result of 200/0 reduction from a two-state model.

More interestingly. the fitted curves clearly show the trends of the relationship between the number of gating charge particles and the channel open threshold. The smaller the number of gating charge particles. the more negative membrane potential is needed to obtain the same channel open-probability. The potential difference between the pre- and post-shock membrane potentials to open 5% (eol) of the K+ channels is about 5 mY. These results clearly indicate that after electrical shock, not only the number of gating charged particles is reduced, but also the channel open threshold shifts a few m V in the negative direction.

o

c:o§. -1

f C1. -2

! ~ m -3~-v--rr---------------------------6

4~-----r----~------r-----.-----~-----'

-60 ·50 40 -30 -20 -10 o


Figure 18. The same data from Figure 17 was analyzed by a multiple-state model of the K+ channels. Equation 7 was used to fit the data. Taken from Chen et a/ .• 1998. b.

Reductions in the channel gating charged particles resulting in both a smaller slope of the open probability-curve and a negative-shift of the channel open-threshold can also be observed

181

from the numerical plots of the channel open-probability as a function of charged particles by Almers and Hille (Almers, 1978, Hille, 1992). They employed a two-state model and further assumed that Eo-Ee equal to zero, therefore half of the channels are in the closed state when no external electrical field was applied. The numerical plots showed that the smaller the limiting number of gating charged particles, the lower the slope of the open probability is, and the more negative membrane potential is needed to get the same channel open probability.

In summary, after the demonstrations that the electric field-induced functional reductions in membrane proteins are mainly not from thermal damages by Joule heating effects, we postulated that the electric field-coupled electroconformational damages may playa significant role in electric injury. According to Boltzmann theory for the channel voltage-dependence, a comparison of the pre- and post-shock channel conductances predicts that possible conformational alterations may be located in the channel gating system expressed as a reduction in the number of limiting gating charged particles.

We further performed direct measurements of the intramembrane charge movement currents of the membrane proteins. The results showed that a brief electric shock may significantly change the characteristics of the charge movement currents, including reductions in the amount of charges, increase in the delay-time of the slow hump component, Qy, and broadening its shape. These studies provide direct evidences to support our hypothesis that high-voltage electric field can cause proteins' electroconformational changes, reducing the amount of movable charged particles, and, therefore, resulting in failures in proteins' functions. These results show that, besides the thermal damages and the electroporation of cell phospholipid bilayer, the electrocoupled structural damages in membrane proteins is an important mechanism involved in electric injury, especially in a field range not high enough to cause thermal damages.

IV. ELECTRO-COUPLING OF WEAK OSCll..LATING ELECTRIC FIELD AND MEMBRANE PROTEINS

1. Theories for Activating Membrane Proteins by an External Electric Field

Since late 1970s, activation of the transport of Rb+ or other ions across erythrocyte membranes mediated by the Na+/I(+ ATPase have been experimentally demonstrated being attributed to the applied regularly oscillating electric field. (Serpersu and Tsong, 1983, 1984; Teissie, 1986; Liu et a/., 1991).

To interpret these experimental results, different theoretical models of transportermediated ligand transports or enzyme-catalyzed reactions under the influence of an alternating electric field have been postulated. (Tsong and Astumian, 1986, 1987; Chen, 1987; Astumian et a/., 1987, 1989; Markin et al., 1990; Markin and Tsong, 1991; Tsong, 1992). Studies in predicting external perturbation enhanced functions in the membrane proteins has also been extensively conducted. Among many publications, some excellent reports include: A noiseinduced enhancement of signal transduction across the voltage-dependent alamethicin channels in artificial membrane (Weaver and Astumian, 1990); the relationship of the electrical field bandwidth and the system equilibrium in interpreting the minimum detectable perturbation in the presence of thermal noise (Bezrukov and Vodyanoy, 1995); direct observation of the rotation of

182

F.-ATPase (Noji et al., 1997); and recently, a theory of non-equilibrium fluctuation based upon Brownian motion in order to interpret macroscopic electric field-induced uphill transport of ions (Astumian, 1997).

In general, for a passive membrane transport system, the steady-state transport direction is dependent on the electrochemical potential difference across the cell membrane. For the ligandmediated membrane transport, if the ligand is not charged, the direction of the ligand transport is only determined by the concentration gradient of the ligand across the cell membrane. If the ligand is charged, the transport direction is dependent, in addition to the ligand concentration gradient, on the electric potential across the cell membrane.

For an active membrane transport system, the transporter proteins transport ligand against its concentration gradient by expenditure of energy either from ATP molecule or the electrochemical potential generated by cotransporters. If the transporter is charged, the active transport ofligand against its concentration gradient may be achieved by exposing the cell into an oscillating electric field. The prerequisite factors are that the transporters are charged and the kinetic reaction rate constants of the system satisfY certain asymmetry conditions (Tsong and Astumian, 1986; Westerhoff et al., 1986; Chen, 1987; Astumian et aI., 1987). That is, the charged and asymmetric transporter system in the membrane can absorb energy from an applied oscillating electric field to pump ligand across the membrane against its concentration gradient (Tsong and Astumian, 1986). It is obviously that to absorb electric energy provided from an oscillating electric field, the enzyme must involve an energy-activated reaction to pump a substrate, either neutral or charged, against its electrochemical potential difference. Another good example is that, instead of pumping ligand across the cell membrane, the enzyme may utilize the electric energy to synthesize ATP molecules.

Support for the hypothesis of oscillating electric field-induced activation of the membrane active transporter were derived from several theories. When the field-induced membrane potential is comparable to the physiological membrane potential, a theory of electroconformational coupling (ECC) considers that electric energy of an oscillating electric field may be absorbed by the charged transporters introducing structural changes in the transporter. The energy absorbed in the conformational changed transporter can be used to drive the chemical reaction away from its equilibrium state.

The foundational reason behind a shift in the steady state due to the presence of an electrical perturbation is the relative shift in energies of the two states involved. If the two conformational states have permanent dipole moments which differ by 1000, the relative energy shift between the two states, due to imposition of a 50 m V transmembrane potential change, is 1.4 kcallmole and the equilibrium constant is changed by a factor often. When exposed to an electrical field an energy called "electroconformational energy" can be stored in the proteins. Removal of the electric field results in a release of this energy to the environment via relaxation to the zero-field equilibrium. In fact enzymes coupled to an electric field is via a cyclic mechanism (Tsongetal., 1987).

Whereas the field-induced membrane potential is far small, not enough to introduce protein structural changes, several theories were postulated to interpret the weak oscillating electric field-generated activation of the living system. Instead of driving away from chemical equilibrium, the reaction rate of a spontaneous chemical reaction is modified by the weak oscillating electric field. Some of these theories include the oscillating activating barrier theory and the theory of surface current compartment. Since not involving proteins conformational changes, there theories will be only briefly described later in this review.

183

1) Theories from Enzymatic Dynamics Studies

As described by the Michaelis-Menten equation, an enzymatic reaction recycle from a substrate state to a product state and then back to the substrate state. In other words, the enzyme molecule microscopically oscillates along its reaction states in the catalytic cycle. The equivalent oscillating frequency is equal to the turnover number.

When a transport molecule is spatially fixed and exposed to an anisotropic medium, like the enzyme inserted in cell membrane, the equilibrium state of the reaction rate depends on the molecular conformation state and the electrochemical potential difference. If the molecule has only two conformational states, El and ~, in equilibrium, and the transition between the two states involves charge particles, the equilibrium will be shift according to an applied electric field. When cells are exposed to an oscillating electric field, the field may enforce these membrane molecules to oscillate microscopically between the two conformational states.

Many membrane proteins, especially, the membrane active transporters, such as the Na+/K+ ATPase, Ca2+ -ATPase and proton pump, are involved in this category. Obviously, if the frequency of an oscillating electric field does fall into the range matching with the turnover frequency of the molecules, there will be intricate molecular interactions with the electric field leading to complex chemical kinetics for the reaction.

In general, ifan enzymatic reaction system has N states, N-I dimension linear differential equations (the kinetic equations) can be written to describe the rate of the whole reaction cycle in terms of the reaction rates in each state.

(9)

The forward and backward reaction rate constants between the ith and jth states are lev and kp respectively. Then, the kinetic equation for this system is

N

dE;/ dt = - L(KijE;- N;EJ), j= 1, ... n (10) ;=1

When an electric potential V is applied to a voltage-sensitive step, the equilibrium constant Kg, which is the ratio of~, must be replaced by K'ij = Kg exp (<lij VIRT), where R is the gas constant and T the absolute temperature. When an oscillating potential V = Vo cos (wt) is applied, the equilibrium constant, K'ij, in the above equation can be replaced by K'ij = K'u exp [(av Qu VoIRT) cos (wt)), where Qu = - qji is the electric charge that moves across the potential V during the i and j transition; and the au = 1-~ is the apportionment constant. This is a nonlinear equation set with coefficients varied periodically. With the Fourier series and the modified Bessel function expansions of the periodical parameters in the equation with up to the second order terms retained, the general solution of the above equation can be obtained. After an integration over a full cycle of the oscillating electric field, the steady-state flux rate, J, of the catalytic cycle can be expressed as (Robertson, 1991):

N-l A; J=JO+L 2 2

; 1+ ,; OJ (11)

184

where, 10 is a constant offset, which is the steady-state flux rate in the absence of the external perturbation. The second term is a summation of Lorentz curves, where "ti is the characteristic time constant and 111:; is the characteristic frequency of each Lorentz curve. A; is the magnitude with either a positive or negative sign. All coefficients of "t's and A's are functions of all of the reaction rates, lev and~, of the whole cycle, even though some rate coefficient are not voltagedependent. The total steady-state enzymatic flux rate, J, is the flux rate in the absence of perturbation, Jo. plus a sum of Lorentz curves which are field frequency-dependent with different magnitudes, signs and characteristic frequencies.

The steady-state flux rate, J, of the enzymatic reaction can be plotted as a function of the electrical field-frequency. Figure 19 simply shows Jo plus two Lorentz forms. The abscissa is the frequency, tll, of the perturbation field and, the ordinate is the enzymatic reaction rate, 1. The plateau marked by an arrow is the reaction rate in the absence of perturbation, Jo. Since A; can have either a positive or negative value, a summation of the two Lorentz curves with opposite magnitudes and close characteristic frequencies will form a bell-shaped curve frequency-dependence. This means that a weak oscillating electric field with a frequency fallen inside the bell-shape can increase the steadystate flux rate of the enzymatic reaction. The locations and the number of the frequency windows are functions of all of the reaction rate in the loop, especially the voltage-dependent steps in the reaction cycle and their reaction rates.

summation

t------ -j - -j -~--.---· i I . · I J • · I ! i

lItz lit,

Frequency

Lorentzian curve 2

Lorentzian curve 1

Figure 19. Summation of two Lorentizian curves as a function of the frequency of oscillating electric field. lh: I and l/'r2 are the characteristic frequencies of each Lorentizian curve.

To achieve the goal of activating proteins by an oscillating electric field, several necessary factors have to be satisfied. One obvious and crucial factor is that an appUed field must retain a certain degree of autonomous charaCteristics. That is, the waveform ofthe appUed field should not be altered by the interaction with the transporter or substrate (Astumian et al., 1987). Another crucial attnoute is that the binding constants of the transporter at cytoplasmic side and extracellular side are difference (Westerhoff et al., 1986; Tsong and Astumian, 1986; Astumian et al., 1989; Chen, 1987; Tsong, 1990). When these requirements are met, the efficiency of energy transduction depends on the fieldinduced membrane potential difference, the voltage-dependent reaction rate, kv and k,.

185

An equivalent approach is to apply an oscillating perturbation and measure one or more oscillating concentrations as a function of frequency (Astumian et al., 1989; Hom, 1993). The amplitude of the oscillation is also a sum of the Lorentzian curves. Another alternative is to measure the power absorbed by the system. Again, the result is a sum of Lorentzian curves, with each Lorentzian amplitudes being quadratic during the perturbation (Tamuraet al., 1984, 1985).

The enzymatic reaction system needs not be electrogenic, but it is required that at least one step in the reaction cycle be sensitive to the membrane potential. In the Na+/K+ pumping reaction loop, steps of occlusion ofNa + ions in the Na + transport limb (Nacao et al., 1989) and possibly the steps of occlusion of K+ ions in the K+ transport limb (Rakowski et al., 1990) have been shown voltage-sensitive.

Tsong et al. has shown that at a frequency of about 1 kHz, oscillating electrical field can activate the Na+/K+ pump in erythrocytes resulting in an increase in the ionic concentration gradient crossing the cell membrane. In our lab we found that with a similar frequency range, a 50 m V oscillating membrane potential can result in a net outward transmembrane current in skeletal muscle fibers. Since all of these effects were eliminated by the application of ouabain, these field-induced activation in cell membrane transport were carried by the Na+/K+ ATPase molecules.

2) Prediction from the Viewpoint of Electric Engineering

The Asymmetric I-V Curve and Rectifier Features. The above theoretical analysis is based on dynamic enzymology study by solving the Michaelis-Menton equations for an enzymatic reaction. The result predicts that an oscillating electric field-perturbation can increase the steady-state flux rate of the enzyme reaction as long as there is one voltage-dependent reaction rate in the loop. On the other hand, from engineering perspective, the experimental results of the asymmetric characteristic of the steady state voltage-current dependence of many electrogenic pump m(jlecules indicates that these molecules, such as the Na+/K+ ATPase, show rectification capability. In other words, for an input of alternating current field, the pump molecules can generate a direct current output.

If we consider the N a + /K+ pump molecule as a hypothetical black box, an input of a symmetric oscillating potential can generate an output of net direct current, the voltage-current dependence of the black box must be asymmetric with respect to the voltage potential bias. Indeed, experimental

results from either myocardial cells (Gadsby, 1984) or skeletal muscle fibers (Chen et al., 1998, d) showed the asymmetric voltage dependence of the Na+/K+pump currents. As shown in Figure 20, the slope for the I-V curve at the potentials below the membrane resting potential is very shallow, the pump current becomes negligible when the membrane potential falls lower than -110 mY. In a potential range from -90 to +20 mY, the slope becomes much steeper. This asymmetric feature of the pump I-V curve is the same as that of an electronic rectifier, where the I-V curve is asymmetric with respect to the potential bias. Similarly, the potential bias for the molecular pumps occurs at the membrane resting potential, -90 mY.

When the cell membrane is held (biased) at its resting potential (the thin-line in Figure 20), an input of symmetrical oscillating electrical field generates an asymmetrical pump current. The positive half cycle that depolarizes the cell membrane can significantly increase the pump current, while the negative half cycle that hyperpolarizes the membrane yields only a small current decrease. An integration of the pump currents with time over a whole stimulation cycle results in a non-zero, net outward pump current. This results indicates that a symmetric oscillating electrical field can be converted by the Na+/K+ pump molecules to a net ion flux across the cell membrane against the ionic

186

concentration gradient. In fact, it is this asymmetric voltage-dependence characteristic that makes the Na+/K+ ATPase molecules capable to maintain a stable membrane resting potential.

;cEo

2S

20

I 15

il a. g 10 a.

! t 5 'll ii

• •

•

• • • • •

O~I---J~~-.-r---r----r---~--~,

-20r -150 .100 -50 50 100


Membraen resting potential (bias potential), ·90 m V

Figure 20. Rectifier-like characteristics of the Na+JK+ pump currents as a function of the membrane potential. Taken from Chen et 01., 1998, d.

One question remains: Under normal conditions, the turnover number of the Na+/K+pump molecules is only 20-70/s, equivalent to a low frequency range. How do the pump molecules respond to a high-frequency electric field? From the I-V curve of the Na+/K+pumps (Figure 20), when the membrane potential depolarized the maximum pump current is many times larger than the current at or below the membrane resting potential. It has been shown that the Na+:K+ coupling ratio is 3:2 and this ratio is insensitive to moderate changes of the membrane voltage (Lauger, 1991). The only explanation is that the depolarized membrane potential accelerates the pumping rate. Moreover, we have observed that the pump molecules retain a rectifier-like voltage-current dependence pot only for the steady state current but also for the transient pump currents. Even for stimulation pulse less than 50 J.lS duration, the evoked Na+/K+ pump currents remain asymmetric characteristics. This result indicates that for a frequency up to 20 KHz, the electric field induced Na+/K+pump currents is a nonzero, net outward current across the cell membrane. All these results clearly show, even at a relative high frequency, that an electrical field should be able to activate the functions ofthe Na+/K+pump.

3) Theories of Low Electric Field-Induced Signal Transduction

The above discussion ofthe electroconformational coupling theory is mainly used to interpret the experimental results where the field-induced membrane potential difference is at the same order of the physiological membrane potential. For example, when an AC field was shown activating effects on the Na+/K+ ATPase, the field-induced membrane potential is about tens ofmillivdlts in the same

187

order of the normal membrane potential. Experimental results also showed an electric field of magnitude of200 mV can effectively facilitate ATPase synthesis in mitochondria membrane. This value is almost the same as the normal membrane potential of mitochondria.

In addition, experimental results have showed that when field strength is far below the physiological value, for example, introducing only a few ~ V across the cell membrane, the electric field can still affect many cellular functions, such as enzyme activity, DNA, RNA, and protein synthesis. A typical example is the weak field-induced ATP synthesis of the F of I-ATPase under a low phosphorylation potential and ATP hydrolysis from rabbit kidney (Blank and Soo, 1989, 1990). For the low field-strength, the electric field can not induce proteins' structural changes, so that the electro-coupling conformational change theory is not able to interpret these experimental results.

In later 1980, Black (1987) proposed a surface compartmental model (SCM), which considers the current produced at the vicinity of the cell membrane as the origin of the field effect. This theory has been well used to interpret their experimental data on the oscillating electric field upon ATP hydrolysis activity ofNa+fI(+ ATPase. Serpersu et al. (1983, 1984) have also successfully used this theory to predict the existence of frequency window of the external electric field.

Tsong et al. postulated the oscillatory barrier (OAB) model to interpret the experimental results of the low field induced activation on the membrane proteins (Markin et aI., 1992, a, b). The basic concept of this model is that the external oscillating electric signal must be able to generate a resonance response of the membrane proteins. Therefore, the fundamental postulate of the OAB mechanism is that the activation barrier of the rate-limiting step of a membrane protein is oscillatory. In fact, there are many modes of motion within the membrane protein structures. Many of these motions involve movement of charged particles or reorientation of the equivalent dipole moments. These electrical movements are inevitable response to the applied oscillating electric field.

2. Experimental Evidences

1) Electric Field-induced Activations in Electrogenic Pump Molecules

Structural features of many membrane proteins susceptible to electric fields have been extensively studied. These include: (I) ionizable groups which lead to different net charges at different pHs, including phosphorylation and de-phosphorylation of proteins (Wada, 1976 ), (2) a-helices, which represent very large dipoles of about 120 D per membrane pass (Wada, 1976; Hoi et al., 1978, 1985), (3) sulfhydryl groups or disulfide bridges which influence the conformational stability of a protein (Eisenberg, 1984). Membrane proteins have structural characteristics which make them sensitive to an external electric field. Their function can, therefore, be regulated by the transmembrane electrical potential.

Teissie and Tsong (1980) measured the ouabain-inhibitable membrane conductance induced by 24 V/cm AC field at 1 kHz. They (1981, a, b) performed a series experiments to monitorNa+, K+ and Rb+ fluxes of AC-treated human erythrocytes by using radioactive tracer. At 4°C, when erythrocytes in isotonic suspension were exposed to an AC field of20 V/cm, a net influx ofK+ or Rb+ was detected at I kHz. When an AC electric field of 1 kHz was used, the maximal influx was found to be at a field strength of20 V/cm. Thus, the AC-stimulated K+ and Rb+ influx exhibited an optimal frequency and an optimal field strength. Under optimal conditions, the net AC-stimulated Na+ pumping rate was 15-30 ions per enzyme, per second and Rb+ pumping rate was 10-20 ions per enzyme, per second.

More interestingly, they found that the AC-stimulated cation pumping apparently did not

188

require hydrolysis of ATP. At 4 °C there was no significantly ATP hydrolysis, but there was stimulated activity. This stimulated activity was almost identical in ATP-containing (600-800 JlM

ATP) and ATP-depleted « 10 JlM ATP) erythrocytes. Because the stimulated transport of cations did not require A TP hydrolysis, the enzyme must be able to absorb free energy from the applied electric field. Only the oscillatory field ofa defined frequency and amplitude was effective. The fieldstrength dependence and the frequency dependence of the Na+ and the Rb+ pump activities are shown

in Figure 21.

~ 60 ::- 30

.r: &;

U 50 U 25 CD CD

B a: a: G; ., 0 40

0 20

E E !!. !!. 30 15 ..

)( :I :::>

20 10 W c

"' 10 .tl 5 Z a: ... ... ID

CD 0 0

2 4 5 6 7 8 0 2 3 4 5 6 7 8

Log (Freq. in Hz) Log (Freq. in Hz)

Figure 21. Frequency dependence of the electric field induced cation pumping by Na+JK+ ATPase of human erythrocytes. A. the Na+ -efflux and B. Rb+ -influx with an AC field of20 V/cm (peak-to-peak) represented by open squares. Taken from Tsong, 1992.

Rephaeli et al. have found that the (NaJ)E1P to ~P2 conformational transition ofNa+JK+ ATPase was 1.67/s in the zero field, but it increased to 6.5/s when Klvalinomycin was used to generate -137 mV across the bilayer (Borgens et al., 1981; Rephaeli et al., 1986).

ATP synthesis under physiological conditions requires 10-15 kcaVmol of energy. In the mitochondrion, this energy is supplied by the electron transport supported by the oxidation offood stuff. The final form of energy before conversion into the y-phosphodiester bond of ATP is the proton electrochemical potential energy according to the Mitchell's hypothesis, or the electric potential energy according to the ECC mechanism. Chauvin, Astumian and Tsong using a pulsed electrical field method observed that the ATP yield of beef heart submitochondrial particles increased with increasing field strength, reaching 8-10 ATP per FoFl complex per electric pulse of 100 Jls when 30KV/em pulses (which can generate a transmembrane potential of400 mY) were used (Chauvin et al., 1987), (Figure 22). This rate is at least 100-fold higher than the turnover rate of solublized enzyme under zero field, and is also much higher than that observed in pH or K-jump experiments. In most of these experiments the electron transport chain of mitochondria was inhibited or the thylaeoids were kept in the dark. Therefore, energy sources from the normal channels were depleted, and any de novo ATP synthesis must derive energy from the applied electric field.

In the low field experiment, Tsong and his colleague have used up to 60 V/cm of alternating

189

electric fields (Ae). ATP yield was low, less than 10.3 ATP per enzyme per cycle ofAC (Liu et 01., 1991). However, because the Joule heating was much less severe than that of the high field experiment, electric stimulation could be continued for a long period of time, e. g., 10 min to 1 h, and the accumulated A TP yield was more than 10 per enzyme.

Calcium pumping by the Ca-ATPase is an important mechanism involved in human erythrocytes. Calcium efflux was found to be stimulated by the AC field. The optimal frequency was found to be 100 Hz, optimal field strength to be 30 V/cm. An optimal ligand concentration for this activity was also found when the Ca2+ concentration is at 0.8 mM. The AC stimulated Ca2+ pumping activity was approximately 15 JiM per liter of cells per hour. The window for the ligand concentration is particularly interesting because the existence of such a window was first predicted by an electroconformational model (Martin and Tsong, 1991).

Z 200r---r---r---~--~~ ~ a ~ 150 o :::!!; ~ (/) 100 iii w I

~ 50 >(/)

el-

A

o

~ OL.. __ ~ __ ~ __ -J ____ L..-~. o 5 1 2 3 4

LOG (FREOUENCY) (Hz)

Z 400

~ a B :::!!; 300 .J 0 • :::!!; ~ (/)

iii w I I- 100 z >-(/)

el-I-<C

25 50 75 APPLIED VOLTAGE (V/em)

Figure 22. ATP synthesis of beef heart submitochondrial particles induced by alternating electric fields. A. Frequency dependence when exposure to a field of 60 v/cm for 10 minutes and B. field-strength dependence. Taken from Tsong, 1992.

A number of cases in which spontaneous regular oscillations occur have also been reported (Yoshikawa et 01., 1986; Chay, 1986; Ringel et 01., 1986), one of the earliest being th~ "Teorell oscillator" (Teorel, 1962). Since these processes cause changes in the local field strength, they must in turn be field-dependent. Thus all such processes are thermodynamically coupled to each other by their common intermediate, the local membrane potential.

Horsthemake and Lefevre (Horsthemake et 01., 1984; Tsong, 1991) studied the role of fluctuations near critical points using the terminology "noise induced phase transitions". In their investigations it was shown that externally imposed noise could trigger a phase transition in which the system would go from one stable state to another. Tsong et 01. have shown that very simple chemical reactions can also be driven from the deterministic equilibrium position by regular oscillations or externally driven random fluctuations ofa thermodynamic parameter (1992). This result indicates that the chemical system has captured energy from the externally enforced fluctuations.

Recently; Tsong and his colleague reported the use of high frequency alternating electric fields

190

to induce deformation of sea urchin eggs, leading to budding of membrane vesicles or fission of cells (Marszalek, 1995). They prepared mini cell bodies from a single egg by carefully manipulating the

frequency and amplitude of the AC field and the ratio between the intere1ectrode spacing and the cell diameter. A medium strength AC field «100V/cm at 2 MHz) was applied to attract the egg to one of the two electrodes via dielectrophoresis. The voltage was then slowly increased to approximately 1000V/cm over approximately 30 seconds. The cell elongated and separated into two fragments. When, the field was turned off. the mother cell and the daughter vesicle retracted to form spherical mini cell bodies. The cell shapes indicate that membranes of these mini cell bodies were not leaky to

ions and small molecules. Directional flow of information and energies is characteristic of many types of biochemical

reactions, for instance, ion transport, energy coupling during ATP synthesis, and muscle contraction. A question arose that if a fluctuating force field, or a noise, can induce such directional flux. Xie et ai. (I997) showed that electric fields which fluctuate both in life time and in amplitude, and thus, better mimicking the transmembrane electric fields of a cell, can also induce Rb + pumping by Na + /K+ ATPase. Increased values for sigma led to a nonmonotonic reduction in pumping efficiency. They found that the calculated efficiency in the energy coupling decreased with increasing sigma value.

Blank and Soo (1989, 1990) have investigated effects of weak AC field (mV per cm), on the ATP hydrolysis activity of Na+/K+ ATPase from rabbit kidney. They have found that an AC current can either stimulate or inhibit the ATP hydrolysis activity of the enzyme depending on the degree of ion activation, i.e., the Na+/K+ ratio. Under normal conditions, the enzyme in AC currents showed a decreased ability to split ATP. Conversely, at lowered activity, the enzyme showed an enhanced ability to split ATP (Blank and Soo, 1989, 1990). Maximal effects of AC were around 100-200 Hz.

Applied electric fields have been shown to stimulate DNA, RNA and protein biosynthesis, cell proliferation (Adey, 1981; Robison, 1985) and other cellular activities (Bassett et ai., 1977; Jaffe et al., 1979; Borgens etal., 1981). Recent advances in molecular biology have prompted investigators to search for the molecular origins of the electrical activities of cells and tissues. How bioelectricity reflects and influences biochemical reactions, and how electric fields are used in signal and energy transduction in biological systems are important and fascinating questions.

In broader terms, the widespread occurrence of endogenous weak currents during embryonic development and the in vitro responses of embryonic cells to applied electrical fields suggest an important role for electrical currents in directing the emergence of structural spatial patterns during development. A few examples of cell types in which this phenomenon has been studied include neurite (patel and Poo, 1982), the myoblast (Hinlle et al., 1981) of Xenopus, the neural crest cells of quail (Erickson and Nuccitelli, 1982) and epithelial cells of xenopus embryo (Luther, 1983).

2) Direct Identification of the Na+Jk+ Pump Currents in Skeletal Muscle Fibers

Most ofthe above experiments showing the field-induced activation ofthe electrogenic pumps are based on the indirect measurements of the pump functions, such as using fluorescent dye and radioactive labeled tracers. None of the experiments was performed by directly and simultaneously monitoring the pump currents during the exposure to an electric field, which makes the investigation of screening the optimal parameters of the field more difficult. In this section, a progress in direct measurement of the Na+/K+pump current in skeletal muscle fibers will be briefly discussed (Chen and Han, 1996; Chen, 1998, e).

Electrogenic current of the Na+/K+pump molecules on myocardial and nerve cells have been well studied (Gadsby et ai., 1984; Rakowski et ai., 1990; De Weer, 1990; Lauger, 1991). However,

191

much less results have been reported to the Na+/K+pump activities in skeletal muscle fibers, although the pump population in the red skeletal muscle fibers is much larger than that in heart and nerve cells. So far, the Na+/K+ pump functions can only be indirectly monitored using fluorescent dye and radioactive labels. The possible reasons are that the skeletal muscle fiber is difficult to be spatially voltage clamped. Not like cardiac cells, the skeletal muscle fiber is too long for a microelectrode to provide an uniform membrane potential; while the fiber diameter is too small to insert an electrode along the fiber as widely used for nerve cells. Recently, by using an improved double Vaseline gap voltage clamp technique, we succeeded in developing a method to directly identifY the Na+/K+pump currents in the skeletal muscle fibers.

The fiber preparation is based on the method developed by Hille and ampbel (1976) which is widely used in many laboratories (Kovacs, 1983; Irving, 1987; Hui and Chen, 1992) to study the intramembrane charge movement currents. The experimental protocol is similar to the protocol used to identify the Na+/K+ pump currents in cardiac muscle fibers (Gadsby, 1984; De Weer, 1990; Rakowski et al., 1990). Ouabain, a specific inhibitor for the Na+JK+ pumps, was used to identify the pump currents.

D=95908000 Disk=5464 Epi ~=-40

100

I.\,.

'r

~ -100

o 100 200

D=95908006 Disk=5464 Epi ~=-40

100

~ -100

r

o 100 200

300 Time(ms)

300 Time (ms)

Figure 23. Transmembrane currents responding to a sequence of stimulation pulse holding the membrane potential from -150 to +60 m V. Upper panel: in the absence ouabain; Lower panel: with 0.1 mM of ouabain.

192

A stimulating pulse sequence consisting of fifteen 200 ms duration pulses was used to hold the membrane potential from -150 m V to +60 m V at an increment step of 15 m V. The elicited transmembrane currents were recorded, shown in the upper panel of Figure 23. The last 30 data points during each stimulation pulse were averaged as the corresponding transmembrane currents and are plotted as a function of the membrane potential. A straight-line was fitted to the current points corresponding to the membrane potentials lower than -110m V. This is based on a fact that pump currents and other non-linear currents are negligible when the membrane is hyperpolarized beyooP this potential (Gadsby, 1984). The fitted straight line was extrapolated up to the membrane potential of 50 m V and, then, subtracted from the recorded currents. The differences are the evoked non-linear transmembrane currents. These non-linear currents, denoted by r.-. pump' include the residual channel currents, the membrane non-linear leakage currents, and the active transport currents including the Na+/K+pump currents.

The external solution in the central pool was then changed to the same solution but with 0.1 mM ouabain. The same stimulation pulse sequence induced transmembrane currents are shown in the lower panel of Figure 23. With the same method, the resultant non-linear transmembrane currents were obtained in the presence of ouabain. These currents, 1.-, are a sum of all ofthe membrane's nonlinear currents except the pump currents.

The non-linear currents in the presence of ouabain, 1,,011' were then subtracted from the nonlinear currents in the absence of ouabain, loon, pump. The final current-difference is the Na+/K+ pump current or the ouabain-sensitive current. Figure 24 shows the pump currents plotted as a steady-state I-V curve (dark cycles). Alternatively, several other methods were used to obtain the non-linear currents: either eliminating K+ ions in the bath solution or removing Na+ ions from the internal solutions. The pump currents obtained by the method of eliminating K+ ions in the bath solution are also showed in Figure 24, represented by open squares.

To further confirm our results, the PIN method, which is widely used in··determining the transmembrane non-linear currents, was modified and employed to identifY the non-linear transmembrane currents. The I-V curves of the N a + /K+ pumps currents are consistent through all these methods.

nA

25

20

15

10

5

o •

o •

o

• o

·5+-----~------r------r-----,._----,_----_.

·200 ·150 ·100 ·50 o 50 100 mV

Figure 24. The NaIK pump currents as a function of the membrane potential. Dark cycles and open squares represent the pump currents obtained by adding 0.1 rnM ouabain and eliminating K+ ions in bath solution, respectively.

193

The voltage-dependence of the pump currents in the skeletal muscle fibers is similar to that obtained from cardiac muscles (Gadsby et 01., 1984). Interestingly, the Na+/I(+ pump currents measured from skeletal fibers are larger than the pump currents obtained from cardiac muscles, which is consistent with the results reported by histology study showing that the density ofNa+/K+ ATPase in red skeletal muscle fibers is higher than that of cardiac muscles (Guyton, 1982).

It is worthwhile to point out that the I-V curve of the Na + /K+ pump molecules has a nonlinear characteristics. The I-V curve is asynunetric with respect to the membrane resting potential at -90 mY, which is similar to the I-V curve ofa semi-conductor rectifier. Therefore, it is clear that the Na+/K+pump molecules have the rectifier-like characteristics.

3) Asymmetric Characteristics of the Na+/K+ Transient Pump Currents

It is necessary to emphasize that the Na+/K+ pump I-V curve with non-linear, rectifier-like features was obtained from the steady-state pump currents. The currents were sampled 200 ms after the changes in the membrane potentials. To study an oscillating electrical field-induced effects on the pump functions, it is necessary to show whether the voltage-dependence of the pump transient currents remains asynunetric with respect to the membrane resting potential. Using our improved double Vaseline gap voltage clamp technique, we can observe the fast response up to 50 IlS after the membrane potential changes (Chen, 1998, b, e).

1000

<-.::. c ~ 0 ::s

(.)

-1000

A. ..

I (.)

0

Transmembrane Currents Responding to Symmetric Pulses

0.8

·IOmV .90mV

·170mV

Internal solution: 40 mM Na; 6 mM A TP External solution: 120 mM Na; 5.4 mM K

Ca)

Time <msl $1Z£807

Transmembrane Current Associated with NaIK Pump Activity

(Ii + I - I )oon"ol- (Ii + 1- i )quabain

(b)

o 0.8 Time <msl Figure 25. Asymmetric characteristics of the transient Na+fI(+ pump currents with respect to the membrane resting potential of -90 mY.

194

To study the transient pump currents, the cell membrane po~ential was held at -90 mY. A sequence of 600 I1S pulse-pairs was applied to the cell membrane. Each pair consisted of two oppositely polarized pulses having a magnitude symmetric with respect to the membrane holding potential, shown as the inset of the upper panel of Figure 25. The evoked transmembrane currents are superimposedly shown in the middle panel. The current corresponding to the positive pulses is denoted as ~ while the current corresponding to the negative pulse in the same pair is denoted as I.j. Subscription "i" represents the i th pair ofthe stimulation pulses. The difference of these two currents, (~ -I.J ..... IIuI, represents the asymmetric features of the cell membranes with respect to the membrane holding potential.

After adding 0.1 rnM ouabain to the external solution, we use the same method to obtain the current-difference, (~ - I.J ........... The Na+/K+ pump currents were then identified by subtracting the transmembrane currents in the presence of ouabain from those in the absence of ouabain. Therefore the asymmetric features of the pump currents can be described as follows:

Ipllmp. arymmetry = (II - I - I )control - (It - I - I )ouaba/n (12)

The results are shown in the lower panel of Figure 25. If the pump currents were symmetric with respect to the membrane holding potential, the resultant current-difference should be zero. These nonzero current-differences of the Na+/K+ pumps clearly show that the pump transient currents remain asymmetric features with respect to the membrane holding potential. Fig. 25 implies that an oscillating membrane potential with a frequency up to tens of kHz can still activate the Na+/K+ pumps to generate a net outward pump current.

4) Direct Measurements of the Electric Field-Induced Na+/IC Pump Current in the Skeletal Muscle Fibers

More recently, we directly studied an oscillating membrane potential-induced activation in the Na + /K+ pump functions (Chen, 1998, b, e). We applied a continuously oscillating membrane potential to the cell membranes and simultaneously monitor the evoked transmembrane currents. Meanwhile, we investigated whether some optimum frequencies existed where an electrical field could activate the Na+/K+ pumps more effectively than other frequencies.

There were three periods of the oscillating membrane potential waveform: a pre-stimulation period holding the membrane at the membrane resting potential for 400 ms; a stimulation period of 40 ms duration oscillating the cell membrane with 50 m V in magnitude and various frequencies from 100 Hz to 10KHz; and a post-stimulation period returning to the membrane resting potential.

The purpose of using the pre- and post-stimulation periods is to determine the base-line current during the stimulation period. This baseline was subtracted from the recorded transmembrane currents during the oscillating period. We then integrated the oscillating currents with time from the tenth cycle to the last fifth cycles of the oscillating membrane potential. Not using the first 10 and last 5 cycles was to avoid the charging and discharging effects during the "on" and "otr' phases of the oscillating stimulation. The net ions transported during each cycle of the oscillating waveform can be represented by the valley points of the integrated curve. One can readily observe a slight increment in the valley point profile when the number of cycles increases. This trend indicates an accumulation of transmembrane ions.

The same oscillating membrane potential waveform was reapplied to the cell membranes after 0.1 mM ouabain was added to the external solution. Subtracting the ion flux obtained in the presence

195

of ouabain from the flux in the absence of ouabain yielded the net ion flux across the cell membrane carried by the Na+/K+ pumps.

Several sinusoidal frequencies have been studied. The net transmembrane ion fluxes carried by the N a + /K+ pump molecules are plotted as a function of the oscillating frequency, shown in Figure 26. The dark triangles represent the net transmembrane ion fluxes without ouabain and the open squares represent the fluxes in the presence of ouabain.

pC 80

60 ..to. ..to. ..to.

..to. ..to. ..to. ..to.

..to. 40

..to. ..to. ..to.

A. ..to. A.

20

0 a a a a a a a a a a a 0 0 0

·204------.------.-----.------.-----. o 2 3 4 5 ms

Figure 26. Net transmembrane ions carried by the Na+JK+ pumps during the oscillating stimulation as a function of the stimulation wavelength. Dark triangles and open squares represent the ion flux in the absence and presence of ouabain, respectively. Taken from Chen et aJ., 1998, d.

These results suggest that an oscillating membrane potential at a frequency close to 1 kHz can generate larger ion flux across the cell membrane against the electrochemical potential than other frequencies. When 0.1 mM ouabain was added to bath solution, this oscillating field-induced, net ion

flux was eliminated. These results clearly show that the oscillating membrane potential-elicited ion flux was carried by the Na + /K+ pump molecules, and there exists an optimal frequency window around

I kHz. In other words, these studies provide strong evidence that the electrogenic pump molecule, such as the Na+/K+ ATPase, can be activated by a symmetric oscillating electric field.

3. Perspective

Many proteins carrying charged particles function as signal transducers and energy transporters in the living system. These proteins are inevitably sensitive to an applied electric field,

especially those in the cell membranes, because of subjecting to the majority voltage drops induced

by the electric field. When the intensity ofthe applied electric field is high enough, the exerted, huge electrical force may alter the structures of the proteins interacted with the charged particles, resulting in protein's functional reductions. This protein'S electroconformational damage has been recently

196

considered as one of the important mechanisms involved in electrical injury. One of the characteristics of the protein's conformational damages, in contrast to thermal damages, is that the degree of the membrane damages mainly depend on the polarity and the magnitude of the induced membrane potential rather than the transmembrane current.

On the other hand, when the frequency and magnitude of an electric field are comparable to the tune of the intrinsic membrane potential, the electrical field may reversibly oscillate the structures of the charged particles in the membrane proteins with a loop-pattern reaction. The electrical energy may be absorbed by the proteins and consumed in the proteins' reaction loop. Therefore the weak oscillating electric field can be used to maintain and even activate the functions of these membrane proteins.

It has been shown that depolarization ofthe membrane resting potential is a common syndrom for many diseases and the side effects of clinical treatments. A fascinating question arose whether it is possible to use an weak, oscillating electric field to activate the molecular pumps in order to restore the normal membrane potential.

Theoretical and experimental studies indicate that a weak oscillating electrical field can increase the net reaction rate of the Na+/I{+ pump molecules. In spite of promising results, more detailed investigations need to be done to answer many important, fundamental questions prior to considering any practiclJ,l application. For example, what is the efficiency of an oscillating electric field-induced activation in the Na+/I{+ pump molecules under the physiological conditions? Can we avoid the electric field-induced effects on other membrane proteins, especially the voltage-sensitive ion channels? Is this field-activated pump function fully independent on the consumption of ATP molecules? What is the efficacy of the field-activated pump functions in saving electrically injured muscles?

Based on the time delay in opening the K+ channels and the quick inactivation of the Na + channels, it is possible that when a frequency is higher than a certain range (a few kHz), these ion channels can not respond to the fast oscillating stimulations. Therefore, investigations in a range of relatively high frequency need to be done.

Furthermore, in addition to the currently used voltage clamp technique in in vitro study to measure the pump current, new methods should be developed in monitoring the pump functions in intact fibers. In contrast to the traditional method using radioactive- and fluorescent-dye labeling to monitor changes in ionic concentration, continuously and directly monitor changes in the membrane potential is an efficient method to study the electric field-induced activation in the pump molecules. This is because the low-capacity energy source carried by electric potential difference, alJl, is built up much faster than the high-capacity energy source, a chemical component of the electrochemical energy.

Finally, based on the results on healthy fibers, oscillating electric fields should be applied to the electrically injured muscle fiber in order to study the efficacy in saving these fibers. Results of these studies may lead to further in vivo investigation and finally to a development of an effective, non-invasive therapeutical technique. This technique is not only beneficial to electrically injured patients, but may also has potential application to many diseases with common characteristic of depolarization in membrane potential.

ACKNOWLEDGMENTS

This research is supported by the grant from National Institute of Health, GM 50785.

197

REFERENCES

Abidor, I.G., Arakelyan, V.B., Chemomordik, L.V., Chizmadzhev, Y.A., Pastushenko, V.F., and Tarasevich, M.R 1979, Electric breakdown of bilayer membranes. I. The main experimental facts and their qualitative significance. Bioelectrochem. Bioenerg. 6:37-52.

Abramov, G.S., Capelli-Schellpfeffer, M., and Lee, RC., 1996, Alteration in sensory nerve functions following electric shock. Bum 22(8):602-606.

Adey, W.R, 1981, Tissue interactions with nonionizing electromagnetic fields, [review], Physiol. Rev. 61:435-514.

Adrian, R.H., and Peres, A.R, 1977, A gating signal for the potassium channel? Nature, 267:800-804.

Almer, W., 1978, Gating currents and charge movements in excitable membranes, Rev. Physiol. Biochem. Pharmacol., 82:96-190.

Alters, R W., 1967, Biochemical aspects of active transport, Ann. Rev. biochem, 36,727-756. Armstrong, C.M., 1981, Sodium Channels and Gating Currents, Physiological Rev. 61(3):644-683. Astumian, R.D., Chock, P.B., Tsong, T.Y., Chen, Y.D., and Westerhoff, H.V., 1987, Proc. Natl.

Acad. Sci. USA, 84:434-438. Astumian, RD., Chock, .p.B., Tsong, T. Y., and Westerhoff, H. V., 1989, Phys. Rev. A39:6416-6435. Astumian, RD., 1994, Electroconformational coupling of membrane proteins, Annals 0/ New York

Academy o/Science, V720: 136-140. Astumian, R.D., 1997, Thermodynamics and kinetics of a Brownian motor, Science, V276:917. Atwater, I., Ribalet, B., and Rojas, E., 1978, J. Physiol. (London) 278:117-139. Atwater, I., Dawson, C.M., Ribalet, B., and Rojas, E., 1979, J. Physiol (London) 288:575-588. Bassett, C.L., Pilla, A.A., and Pawluk, Rl, 1977, A non-operative salvage ofsurgicaUy-resistant

pseudarthroses and non-unions by pulsing electromagnetic fields. A preliminary report, Clin. Orthop. ReI. Res. 124:128-142.

Benz, R, and Zimmerman, U., 1980, Relaxation studies on cell membranes and lipid bilayers in the high electric field range. Bioelectrochem. Bioenerg. 7:723-739.

Benz, R, and Zimmerman, u., 1981, The resealing process oflipid bilayers after reversible electrical breakdown. Biochim. Biophys. Acta. 640, 169-178.

Bezrukov, S.M., and Vodyanoy, I., 1995, Noise-induced enhancement of signal transduction across voltage-dependent ion channels, Nature, V378:362-364.

Bingham, H., 1986, Electrical bums, Clin. Plast. Surg. 13:75. Blanchard, S.M., Damiano, Rl, Smith, W.M., Ideker, RE., and Lowe, lE., 1989, Interpolating

unipolar epicardial potentials from electrodes separated by increasing distances. Pace. Pacing Clinical Electrophysiology. 12(12): 1938-1955.

Blank, M., 1987, Biochim, Biophys. Acta, 902:277-294. Blank, M. and Soo, L., 1989, Bioelectrochem. Bioenerg. 22:313-322. Blank, M. and Soo, L., 1990, Bioelectrochem. Bioenerg. 24:56-61. Borgens, RB., Roederer, E., and Cohen, M.l, 1981, Enhanced spinal cord regeneration in lamprey

by applied electric fields, Science, Washington, 213:611-627. Catterall, W.A., 1988, Structure and function of voltage-sensitive ion channels, Science 242:50-61. Chandler, W.K., and Hui, C.S., 1990, Membrane capacitance in frog cut twitch fibers mounted in a

double Vaseline-gap chamber, J. Gen. Physiology, 96:225-256. Chang, D.C. and Reese, T.S., 1990, Changes in membrane structure induced by electroporation as

revealed by rapid-freezing electron microscopy. Biophys. J. 58, 1-12.

198

Chang, D.C., Gao, P.Q., and Maxwell, B.L. 1991, High efficiency gene transfection by electroporation using a radio-frequency electric field. Biochim. Biophys. Acta, 1992, 153-160.

Chauvin, F., Sstumian, RD., and Tsong, T.Y, 1987, Voltage sensing of mitochondrial ATPase in pulsed electric field induced ATP synthesis, Biophys. J. 51:243a.

Chay, T.R., and Keizer, J., 1983, Biophysical J. 42:181-190. Chay, T.R., 1986, In Nonlinear oscillations in biology and Chemistry, ed. by H.G. Othmer,

Springer-Verlag, New York. Chen, YD., 1987, Proc. Natl. Acad Sci. USA, 84:729-733. Chen, W., and Hui, C.S., 1992, Existence ofQy in frog cut twitch fibers with little QJI. Biophy. J.,

59:503-507. Chen, W., Li, P. and Lee, R.C., 1992, Efficacy of non-ionic surfactants for sealing

electropermeabilized skeletal muscle fibers. Biophysical Journal. V61(2):A422. Chen, W., and Lee, RC., 1994, a, An improved double Vaseline voltage clamp to study

electroporated skeletal muscle fibers, Biophys. J. 66:700-709. Chen, W., and Lee, RC., 1994, b, Altered ion channel conductance and ionic selectivity induced by

the imposed membrane potential pulse, Biophysical Journal, 67:603-612. Chen, W., and Lee, RC.; 1994, c. Electromediated permeabiIization of frog skeletal muscle cell

membrane: Effect of voltage-gated ion channels, bioelectrochemistry and Bioenergetics, 34:157-167.

Chen, W., and Lee, R.C., 1994, d. Evidence for electrical shock-induced conformational damage of voltage-gated ionic channels, Annals oj New York Acad. Sci., 720: 124-13 5.

Chen, W., Han, Y, and Lee, Re., 1996, Effects ofsurpramembrane potential on charge movement currents on frog skeletal twitch muscle fibers, Biophysical Journal, 70(2):A235.

Chen, W., and Han, Y, 1996, Voltage-dependence ofNaIK pump in isolated frog skeletal muscle membranes. Biophysical Journal, 70(2):A18.

Chen, W., Han, Y., and Lee, RC., 1997, Effects of high intensive electrical field on change movement currents on transverse tubular membrane of frog skeletal muscle fibers, Biophysical Journal, 71(2): A.

Chen, W., Han, Y., Chen, Y, and Astumian, D.R., 1998, a, Electric field-induced functional reductions in the K+ channels mainly resulted from supramembrane potential-mediated electroconformational changes, Biophysical Journal, 75: 196-206.

Chen, W., Han, Y, Chen, Y, and Xie, J. T., 1998, b, Field-induced electroconformatiQnal damages in cell membrane proteins: A new mechanism involved in electrical injury. Bioelectrochemistry and Bioenergetics, 47:237-245.

Chen, W., Han, Y, Chen, Y, and Lee, RC., 1998, c, Supramembrane potential-induced electroconformational changes of the voltage-gated Na channels is a possible mechanism in electrical injury, Bioelectrochemistry and Bioenergetics, (in press).

Chen, W., Han, Y, and Chen, Y, 1998, d, Activating the electrogenic molecular pumps, NaIK ATPases, by weak oscillating electrical fields, Bioelectrochemistry and Bioenergetics. (In press).

Chen, W., and Chen, Y., 1999, a, Proteins' electroconformational changes: I. Reduction in the gating charge particles of the K+ channels in frog skeletal muscles, Journal oj PhySiology, (submitted).

Chen, W., Chen, Y., and Han, Y., 1999, b, Proteins' electroconformational changes: II. Reduction in the charge movements offrog skeletal muscle fibers, JournalojPhysiology, (submitted).

199

Chen, W., 1999, c, Oscillating electrical field induced activation of the Na/K pump molecules, (in preparation).

Chernomordik, L. V., Sukharev, S.I., Adidor, I.G., and Chizmadzhev, Y.A. 1983, Breakdown of lipid bilayer membranes in an electric field. Biochim. hiophys. Acta. 736,203-213.

Chernomordik, L.V., Sukharev, S.I., Popov, S.V., Pastushenko, V.F., Sokirko, AV., Abidor, I.G., and Chizmadzhev, Y.A, 1987, The electrical breakdown of cell and lipid membranes: the similarity of phenomenologies. Biochimica et Biophysica Acta. 902:360-373.

Chilbert, M., Mainman, D.A, Sances, AJr., 1985, Measure of tissue resistivity in experimental electrical bums, J. Trauma. 25(3):209.

Cole, K.S., 1972, Membrane, Ions and Impules." University of California Press, Berkeley. DeMello, W.C., 1982, Cell-to-cell communication in heart and other tissues. Prog. Biophys. Mol

Bioi. 39:117-182. DeWeer, P., Gadsby, D.C., and Rakowski, RF., 1988, Annu. Rev. Physiol. 50:225-242. De Weer, P., 1990, The Na/K pump: A current-generating enzyme. In Regulation of Potassium

Transport across Biological Membranes, ed. Reuss, L., Szabo, G., and Russel, lM., University of Texas Press, p5-22.

Dillon, S.M., 1991. Optical recordings in the rabbit heart show that defibrillation strength shocks prolong the duration of depolarization and the refractory period. Circulation Research. 69:842-856.

Dillon, S.M., 1992. Synchronized repolarization after defibrillation shocks, A possible component of the defibrillation process demonstrated by optical recordings in rabbit heart. Circulation Research. 85:1865-1878.

Dimitrov, D.S., Sowers, AE., 1990, Biochemica et Biophysica Acta 1022:381-392. DiVincenti, F.C., Moncrief, lA, and Pruitt, Jr. B.A., 1969, Electrical injuries: a review of65 cases,

J. Trauma. 9:497. Downar, E., Janse, M.J., and Durreer, D., 1977, The effect of acute coronary occlusion on

subepicardial transmembrane potentials in the intact porcine heart, Circulation, 56:217. Ehrenberg, B., Farkas, D.L., Fiuhier, E.N., Lojewska, Z. and Loew, L.M. 1987. Membrane potential

induced by external electric field pulses can be followed with a potentiometric dye. Biophys. J. 51:833-837.

Eisenberg, D., 1984, Three-dimensional structure of membrane and surface proteins, A Rev. Biochem. 53:595-623.

Erickson, C.A., and Nuccitelli, R, 1982. Embryonic cell motility can be guided by weak electric fields. J. Cell Bioi. 95(2. Pt.2):314a.

Gadsby, D.C., 1984, The Na/K pump of cardiac cells, Annu.Rev. Biophys. Bioeng. 13:373-398. Gin, N. J., Department of Physiology and Biophysics, Indiana University, personal communication. Glaser, RW., Leikin, S.L., Chernomordik, L.V., Pastushenko, V.F., and Sokirko, AI. 1988.

Reversible electrical breakdown oflipid bilayers: formation and evolution of pores. Biochim. Biophys. Acta. 940:275-287.

Gourbiere, G., 1992, Work-related electrical bums, 14th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Oct. 29-Nov. 1, Paris, France.

Gross, D., 1988, Electromobile surface charge alters membrane potential changes induced by applied electrical fields. Biophys. J. 54:879-884.

Gross, D., Leow, L.M. and Webb, w.w., 1986, Optical imaging of cell membrane potential changes induced by applied electric field. Biophys. J. 50:339-348.

200

Guy, H.R. in: Molecular biology of ion channels: Current topics in Membrane transport, W.Agnew, ed, Academic Press.

Guyton, A.C., 1981, Medical Physiology, W.B. Saunders Co. Philadelphia, London, Toronto. Hibino, M., Shigemori, M., Itoh, H., Nagayama, K, and Kinosita, KJr., 1991, Membrane

conductance of an electroporated cell analyzed by submicrosecond imaging of transmembrane potential, Biophy. J. 59.209-220.

Hille, B., and Campbell. D.T., 1976. An improved Vaseline gap voltage clamp for skeletal muscle fibers. J. Gen. Physiol. 67:265-293.

Hille, H., Ionic Channels of Excitable Membrane, 1992, Sinauer Associates Inc., Sunderland, Massachusetts.

Hinkle, L.,McCaig, C.D., and Robinson, KR, 1981. The direction of growth of differentiating neurones and myoblasts from frog embryps in an applied electric field. J. Physiol. (Lond.). 314:121-135.

HoI, W.G.J., van Duninen, P.T. and Berendsen, H.J.C., 1978, The alpha-helix dipole and the properties of proteins, Nature, Lond. 273:443-446.

HoI, W.G.J., 1985, Effects of the alpha-helix dipole upon the functioning and structure of proteins and peptides, Adv. Biophys. 19:133-165.

Hore, L., 1993, A novel method for the observation of membrane transportation dynamics, Biophysical Journal, 64:281-289.

Horsthemake, W., and Lefevre, R, 1984, In Noise Induced Transitions, Theory and Applications in Physics, Chemistry and Biology, Springer Series in Synergetics, Vol:15, Springer, Berlin.

Huang, C.L-H., 1982, Pharmacological sparation of charge movement components in frog skeletal muscle, J. Physiol. 324:375-387

Hui, C.S., and Chen, W., 1992, Separation ofQII and Qy charge components in frog cut twitch fibers with tetracaine, J. G. Physiol, 99:985-1016.

Hui, C.S., and Chen, W., 1997, Charge movement in cut twitch fibers of rana temporaria containing 0.1 mMEGTA,J. ojPhysiol., 503.3:563-570.

Hunt, lL., 1992, Soft tissue patterns in acute electric burns, in:"Electrical Trauma: the Pathophysiology, Manifestations, and Clinical Management", Lee, RC., Cravalho, E.G. and Burke, J.F., eds., Cambridge University Press.

Irving, M., Maylie, J., Sizto, N.L., and Chandler, W.K., 1987, Intrinsic optical and passive electrical properties of cut frog twitch fibers. J oj General PhySiology, 89, 1-40.

Jaffe, L.F., and Poo, M.M., 1979, Role of subdermal current shunts in the failure of frogs to regenerate, J. Exp. Zool. 209: 115-128.

Jones, J.L., and Jones, R.E., 1982, Effects of tetrodotoxin and verapamil on the prolonged depolarization produced by high-intensity electric field stimulation in cultured myocardial. cells. Federation Proc. 41:1383.

Jones, J.L., Jones, RE., and Balasky, G., 1987, Microlesion formation in myocardial cells by highintensity electric field stimulation, Am. J. Physiol. 253 :H480-486.

Kiline, R, and Fozzard, H.A., 1984, Extracelular [K+] fluctuations in voltage-clamped canine cardiac Purkinje fibers, Biophysical Journal, 46:663.

Kinosita, K., and Tsong, T.Y., 1977, Voltage induced pore formation and hemolysis of human erythrocyte membranes, Biochim. Biophys. Acta, 471 :227-242.

Knisley, S.B., Smith, W.M., and ·Ideker, RE., 1992, Effect of field stimulation on cellular repolarization in rabbit myocardium, Implications for reentry induction. CirculationResearck 70:707-715.

201

Knisley, S.B., Maruyama, T., and Buchana, W.B.Jr., 1991, Interstitial potential during propagation

in bathed ventricular muscle. Biophysical J. 59:509-515. Kovacs, L., Rios, E., and Schneider, M. F., 1983, Measurement and modification offree calcium

transients in frog skeletal muscle fibers by a meta11ochromic indicator dye. J of Physiology,

343, 161-196. Krassowska, W., Frazier, D.W., Pilkington, T.C., and Ideker, R.E., 1990. Potential distribution in

three-dimensional periodic myocardium---Part II: Application to extrace11ular stimulation. IEEE Transactions on Biomedical Engineering, 37(3):267-284

Lauger, P., 1991, In Electrogenic Ion Pumps, Sinauer Associates, Inc. Sunderland, Massachusetts.

Chapter 8, ppI68-225. Lechleiter, 1., Girard, S., Perlta, E., and Clapham, D., 1991, Science, 252: 123-126. Lee, R.C., and Kolodncy, M.S., 1987, Electrical injury mechanisms:dynamics of the thermal

response, Plas. Surg. 80:663. Lee, RC., Gaylor, D.G., Bhatt, D.L. and Israel, D.A., 1988, Role ofce11 membrane rupture in the

pathogenesis of electrical trauma, J. Surg. Res. 44(6): 709. Lee, RC., River, L.P., Pan, F.S., Li, J., and Wo11mann, RL., 1992, Surfactant-induced sealing of

electropermeabilized skeletal muscle membrane in vivo,Proc. Natl. Acad Sci. USA, 89:4524-4528.

Liu, D.S., Astumian, R.D., and Tsong, T.Y., 1991, J. Bioi. Chern. 265:7260-7267. Luther, P.W., Peng, H.B., and Lin, J.J-C., 1983. Changes in ce11 shape and actin distribution induced

by constant electric fields. Na/ure (Lond.).303:61-64. Markin, V.S., Tsong, T.Y., Astumian, RD., and Robertson, B., 1990,J. Chern. Phys. 93:5062-5066 Markin, V.S., and Tsong, T.Y., 1991, BiophysicalJournal, 59:1308-1317. Markin, V.S., Liu, D.S., Rosenberg, M.D., and Tsong, T.Y., 1992, Resonance transduction oflow

level periodic signals by an enzyme: an oscillatory activation barrier model, Biophysical, Journal,61(4):\o45-1049.

Markin, V.S., Liu, D.S., Gimsa, J., Strobel, R, Rosenberg, M.D., and Tsong, T.Y., 1992, Ion channel enzyme in an oscillating electric field. J. Membrane Biology, 126(2):137-145.

Marszalek, P., and Tsong, T.Y., 1995, Ce11 fusion and formation of mini ce11 bodies by high frequency alternating electric field, Biophysical Journal, 68: 1218-1221.

Mehrle, W., Zimmermann, u., and Hampp, R., 1985, Evidence for asymmetrical uptake of fluorescent dyes through electro-permeabilized membranes of Avena mesophy11 portoplasts, FEBS Lett. 185,89-94.

Miles, W.M., and Zipes, D.P. 1991, Cardioversion and defibrillation: clinical aspects. In: Cardiac pacing and Electrophysilogy. edited by N. EI-sherif and P. Samet, Philadelphia, P A: Saunders, P727-736.

Mir, L.M., Orlowski, S., Belehradek, 1., and Paoletti, C., 1991, Electrochemotherapy: potentiation of antitumor effect of bleomycin by local electric pulses, Eur. J. Cancer, 27,68.

Nakao, M., and Gadsby, D.C., 1989, [Na] and [K] dependence of the Na/k: pump current-voltage relationship in guinea pig ventricular myocytes, J. Gen. Physioi. 94:539-565.

Neumann, E., Sowers, A.E., and Jordan, C.A. (Ed.) 1989, "Electroporation and Electrofusion in ce11 biology."Plenum Press, New York.

Noji, H., Yasuda, R., Yoshida, M., and Kinosita Jr, K., 1997, Direct observation of the rotation of FI-ATPase Nature, V386:299.

O'Neill and Tung. L., 1991. Ce11-attached patch clamp study of the electropermeabilization of amphibian cardiac ce11s, Biophy. J. 59: 1028-1039.

202

Pape, P.C., Jong, D.S., and Chandler, WK, 1996, A slow component of intramembranous charge movement during sarcoplasmic reticulum calcium release in frog cut muscle fibers, J. Gen. Physiol., 107:79-101.

Patel, N., and Poo, M.M., 1982, Orientation of neurite growth by extracellular electric fields. J. Neurosci. 2:483-496.

Patton, IN., Allen, lD., and Pantridge, IF., 1984, The effects of shock energy, propanolol, and verapamil on cardiac damage caused by transthoracic countershock. Circulation 69:357-368.

Plonsey, R., and Barr, R.C., 1985, Effect of microscopic and macroscopic discontinuities on the response of cardiac tissue to defibrillating (stimulating) currents. Medical & Biological Engineering & Computing, March: l30-l36.

Prausnitz, M.R, Bose, V.B., Langer, R, and Weaver, lC., 1993, Electroporation of mammalian skin: a mechanism to enhance transdermal drug delivery, Proc. Natl. Acad. Sci. U.S.A, 90, 10504.

Rakowski, RF., Gadsby, D.C., and De Weer, P., 1990, Stoichiometry and voltage dependence ofthe sodium pump in voltage-clamped, internally dialyzed squid giant axon, J. Gen. Physiol. 93:903-941.

Rephaeli, A., Richards, D., and Katlish, SJ.D., 1986, Conformational transitions in fluoresceinlabeled (N a, K) -ATPase in reconstituted into phospholipid vesicles, J. Bio. Chem. 261 :6248-6254.

Ringel, l, asnd Lee, Y.S., 1986, In Nonlinear oscillations in biology arid Chemistry, ed. By H.G. Othmer, Springer-Verlag, New York.

Rios, E., and Brum, G., 1987, Involvement of dihydropyridine receptors in excitation-contraction coupling in skeletal muscle, Nature, 325: 717-720.

Robert, D.E., and Scher, AM., 1982, Effect of tissue anisotropy on extracellular potential fields in canine myocardium in situ. Circulation Research. 50:342-351.

Robertson, B., and Astumian, D., 1991, Frequency dependence of catalyzed reactions in a weak oscillating field, 1991, J. Chem. Phys. 94(11):7414-7418.

Robinson, K.R, 1985 The responses of cells to electrical fields: A review, J. Cell Bioi. 101:2023-2027.

Rossignol, D. P., Decker, G.L., Lennarz, W.J., Tsong, R.Y., and Teissie, l, 1983, Introduction of calcium-dependent, localized cortical granule breakdown in sea-urchin eggs by voltage pulsation, Biochim. biophys. Acta 148, 781-788.

Rouge, RG., and Dimick, AR, 1978, The treatment of electrical injury compared to bum injury: a review of pathophysiology and comparison of patient inanagenient protocols, J. Trauma. 18:43.

Schneider, M.F., and Chandler, W.K., 1973, Voltage dependent charge movement in skeletal muscle: a possible step in excitation-contraction coupling, Nature,

Serpersu, E.H., and Tsong, T.Y., 1983, J. Membrane Biology, 74:191-201. Serpersu, E.H., and Tsong, T.Y., 1984, J. Bioi. Chem. 259:7155-7162. Sharma, V., Stebe, K., Murphy, lC., and Tung, L., 1996, Poloxamer 188 decreases susceptibility of

artificial lipid membrane to electroporation, Biophys. J. 71(6):3229-3241. Sowers, AE., 1988, Fusion events and nonfusion contents mixing events induced in erythrocyte

ghosts by an electric pulse. Biophy. J. 54,619-626.

Stasheff, S.F., and Wilson, W.A, 1992, Axon terminal hyperexcitability seen in epileptogensis in virto, In Ion Channel, ed. By Narahashi, T., Plenum Press Neewe York and London, ppl37-157.

203

Tamura, K, and Schelly, ZA, 1984, J. Phys. Chem. 84:2996. Tamura, K, and Schelly, ZA, 1985, J. Chem. Phys. 83:4534. Tanabe, T., Beam, KG., Powell, lA., and Numa, S., 1990, Restoration of excitation-contraction

coupling and slow calcium current in dysgenic muscle by dihydropyridine receptor complementary DNA, Nature, 336: 134-139.

Teissie, l, and Tsong, T.Y., 1980, J. Membrane Biology. 55:133-140. Teissie, J., and Tsong, T.Y., 1981, a, Biochemistry, 20:1548-1554. Teissie, l, Knox, B., Tsong, T.Y., and Wehele, l, 1981, b, Proc. Natl. Acad Sci. U.SA, 78:7473-

7477. Teissie, l, 1986, Biochemistry, 25:368-373. Telde, E., Astumian, R.D., and Chock, P.B., 1990, Electro-permeabilization of cell membranes: effect

of the resting membrdne potential.1Jiochem. Biophy. Res. Commun. 172(1):282-287. Tomchik, Kl, and Devreotes, P.N., 1981, Science, 212:443-446. Tovar, 0., and Tung, L., 1992, Electroporation and recovery of cardiac cell membrane with

rectangular voltage pulse, Am, J. Physiol. 92:HI128-1136. Tropea, B.I., and Lee, Re., 1992, Analysis of heat injury to the upper extremity of electrical shock

victims, in: "Electrical Trauma: the Pathophysiology, Manifestations, and Clinical Management", ed. Lee,RC., Cravalho, E.G. and Burke, IF.,Cambridge University Press.

Tseng, G.T., Robinson, R.B., Hoffinan, B.F., 1987, Passive properties and membranecurrents of canine ventricular myocytes, Journal General Physiology, 90:671.

Tsong, T.Y., and Astumian, RD., 1986, Bioelectrochem. Bioenerg. 15:457-476. Tsong, T.Y., and Astumian, RD., 1987, Electroconformational coupling and membrane protein

function, Prog. Biophys. Molec. Bioi. 50:1-45. Tsong, T.Y., 1990, Annu. Rev. Biophys. Biophys. Chem. 19:83-106. Tsong, T.Y., 1991, Electroporation of cell membrane. Biophys. J. 60:297-3tl6. Tsong, T., 1992, Molecular recognition and processing of periodic signals in cells: Study of

activation of membrane A TPases by alternating electric fields, Biochimica et Biophysica Acta, 1113:53-70.

Tung, L. and Borderies, lR, 1992, Analysis of electric field stimulation of single cardiac muscle cells. Biophys. J. 63(2):371-386.

Ursell, P.e., Gardener, P.I., and Albalba, A., 1985, Structural and electrophysiological changes in the epicardial border zone of canine myocardial infarcts during infarct healing, Circ, Res. 56:436.

Wada, A., 1976, The alpha-helix as an electric macro-dipole, Adv. Biohpys. 9:1-63. Weaver, le., and Astumian, RD., 1990, Science, 247: 459, Weidmann, S., 1970, Electrical constants of trabecular muscle fammalian heart. J. Physiology.

210:1041-1054. Westerhoff, H.V., Tsong, T.Y., chock, P.B., Chen, Y.D., and Astumian, RD., 1986, Proc. Natl.

Aca. Sci. USA, 83:5734-5738. Wang, T.K, and Neumann, E., 1982, Electric field mediated gene transfer. Biochem. Biophys. Res.

Commun. 107,584-587. Winegar, RA., Phillips, lW., Yongblom, lH., and Morgan, W.F., 1989, Cell electroporation is a

highly efficient method for introducing restriction endonucleases into cells. Mutat. Res. 225:49-53.

204

Xie, T.D., Chen, Y., Marszalek, P., and Tsong, T.Y., 1997, Fluctuation-driven directional flow in biochemical cycle: further study of electric activation ofNa/K pumps, Biophysical Journal, 72(6):2496-2502.

Yoshikawa, E., Omochi, T., and Matsubara, Y., 1986, Biophys. Chem. 23:211-214. Zimmermann, U., Vienken, J., and Pilwat, G., 1980, Development of drug carrier systems: Electric

field induced effects in cell membranes, J. Electroanal. Chem. 116:553-574. Zhou, X.H., Knisley, S.B., Wolf, P.O., Rollins, D.L., Smith, W.M.,and Ideker, R.E., 1991,

Prolongation ofrepolarization time by electric field stimulation with monophasic and biphasic shocks in open-chest dogs. Circulation Research. 68: 1761-1767.

205

BIOLOGICAL EFFECTS OF mGH PEAK POWER RADIO FREQUENCY PULSES

Shin-Tsu Lu and John o. de Lorge

McKessonHBOC BioServices U.S. Army Medical Research Detachment Microwave Bioeffects Branch Brooks Air Force Base, TX 78235

INTRODUCTION

During the early years of research on the potential hazards of Radio Frequency (RF) radiation, scientists have considered the possibility that a pulsed RF field could produce effects other than those produced by continuous-wave (CW) radiation at the same average power. Recent developments in high peak power pulsed microwave systems have rekindled interest in exploring potential biological effects ofhigh peak power radiofrequency pulses. Systems capable of delivering 100 gigawatt (GW) pulses to a transmitting antenna and establishing· hundreds of kV/m peak: E-field intensity in the beam path are known to exist [Agee et al. 1995]. Safety aspects of high peak power, ultra-short electromagnetic signals have been questioned [Albanese et al. 1994] and debated [Merritt et al. 1995, Adair 1995]. Postow and Swicord [1996] concluded that pulsed microwaves produced by radar transmitters could not cause biological effects other t,han those caused by exposure to continuous wave (CW) radiation at the same average power density of the pulsed microwaves. Apparently, their conclusion should not be generalized to include the specific and replicable auditory effects of microwave pulses. Lin [1989] found that few protection guidelines and exposure standards published by various organizations have established limits to guard against potential hazards of pulsed RF radiation. Studies on comparisons of biological effects between high peak: power but low average pulsed RF radiation and CW radiation of equal average intensity (absorption) are sparse. As a result, existing personnel protection guidelines for high peak power pulsed radiofrequency radiation have often been based on guess work with many uncertainties. This lack of an adequate or appropriate biological database for pulsed RF has recently been recognized by at least one standards' setting group, ICNIRP [ICNIRP 1998].

The majority of contemporary personal protection guidelines utilize frequency dependent time-averaged permissible exposure intensities derived from a time-averaged specific absorption rate (SAR). For RF pulses, time averaging involves integration of individual pulses (peak power

Advances in Electromagnetic Fields in living Systems, Vol. 3 Edited by J.e. Lin, KIuwer AcademicJPlenum Publishers, 2000 207

density, pulse width, rise time, fall time, waveform, power spectrum) and modulation characteristics (repetition rate of pulse train or bursts of pulse train). For simplicity, a time

average intensity of square wave modulated RF pulses is the product of peak power density and

duty factor, the ratio of pulse duration to the pulse period of a periodic pulse train (equal to the product of pulse width and pulse repetition rate). Therefore, peak power density can increase inversely with duty cycle or pulse duration and pulse repetition rate. To prevent unintentionally high exposure and to preclude high specific absorption (SA) for decreasingly shorter pulse

durations ofRF pulses, peak power limits are needed. Due to uncertainties, various organizations approach the issue of peak power limits for RF guidelines differently. For example, the Czechoslovakian government lowers the time averaged power density of pulsed microwaves by a factor of 2.5 from that of the CW microwave [reviewed by Michaelson and Lin 1987]. More recently, IEEE [IEEE 1999] recommended the reduction of the time averaged maximum permissible exposure (MPE) by a factor of 5 under pulsed conditions when pulse duration is less than 100 ms. In addition, IEEE also recommended a 100 kVlm peak electric field intensity limit (-26.5 MW/m2 or 2.65 MW/cm 2). ICNIRP [ICNIRP 1998] recommended lowering the MPE

by a factor of 2 for pulsed microwaves. Additionally, ICNIRP further recommended SA limits of 10 and 2 mJ kg-1 for occupational and general public to prevent the occurrence of a

microwave-induced auditory effect. Pulsed RF may cause a specific biological effect, or it may cause enhancement, attenuation, or no modification at all of the biological effect caused by CW RF at an equal average SAR. Knowledge of specific and synergistic effects (positive, negative, or none) of pulse modulation are needed for a valid evaluation of the MPE modifier used in RF pulses.

Analysis of biological effects and health implications of high peak power RF pulses is further complicated by the method of generating RF pulses. Conventional radar is operated in a narrow band mode in which fractional bandwidth is less than 1% of the center (carrier) frequency. Recent development of Ultra-Wi de-Band (UWB) radar has increased the fractional bandwidth to more than 25% [Taylor 1995]. The UWB radar has also been described as carrierless radar. As the name implies, a center frequency is not found. In this type of radar, risetime and duration of the pulse determine the highest and median frequency [Foster 1995]. The lowest frequency of a UWB pulse is determined by the cutoff frequency of the transmitting antenna. With risetime in 100-300 picoseconds (ps) range and pulse duration in one to several nanoseconds (ns) range, the frequency spectrum (90 or 99% of the pulse energy) of the Fourier transform of a radiating UWB pulse can span continuously from megahertz (MHz) to gigahertz (GHz) (Fig. 1). This type ofRF pulse contains a broad spectrum of frequencies such that the interaction with tissue would be expected to be similar to that of both high- and low-frequency narrow band RF. In essence, analyses of biological effects involves the process of synthesizing these combined effects due to multiple frequency exposure and assigning a weighing factor to

each individual frequency component. The electromagnetic pulse (EMP) produced by nuclear detonation possesses a similar problem as a carrierless pulse with a risetime of 10 to 100 ns and pulse width of 400 ns to a few microseconds (Jis) range [Lin 1975]. The EMP produced by,a lightening discharge typically has a duration of peak current from 10 to 100 fJ-S. The synthesizing

procedure for evaluation can be rather tedious, time consuming and full of uncertainties. In addition, biological effects of multi-frequency exposure are virtually unexplored. Another approach to the subject of biological effects of carrierless RF pulses is to extrapolate from the results of animal experimentation with these pulses. The latter evaluation procedure will be the

208

focus of the present review in dealing with the topic of biological effects of high peak power carrierless pulses.

100 I I I I I r 1 I I I I

90 ---+-----1-----+-----1------r-----+-----t-----+-----t-----+-----t----I I I I I I I I r I I

E 80

? 70

~ 60

in 50 Z w 40 I-~ 30 0 --' 20 w ;:;: 10

I w 0

I I I I I I I I I I I ----t------.. ------I"-- ------1"------1------+------+"-----+------ ... -----1'----I I I I I I I I I 1 I I I I I I I I I I ---+-----1-----+-- - -----r-----+-----t-----+-----t-----+-----t----I I I I I I I I I I I I I I I I I I I I I I

----1-------+------1"-- ---f- ---1-------4------+----- ........ -----+------1------ .. ----I I I I I I I I I I I

---+-----~------~-- --~-- ---~-----~------~-----+-----+------~-----~----I I I I I I I I I I I I I I I I I I I I I I ----I-------+------t--- ---+-- --I-------1------+-----+-----+------!-------+----I J I I I I I I I I I I 1 I I I I , I I I I

---+-----t-----+-- --t--- -r-----+-----t-----+-----t-----+-----t----I I I I 1 I I I I I I I I I I I I I I I I I ----1------4------1-- __ -t____ t--____ -I ____ .... _____ -+- _____ + ______ 1-_____ ... ___ _ I I I I I I I I I

____ ~----_~----_-~- ---4------r-----+--- I - --t----- I ---t----I I I I I I I I I I I

-f.: --i:--~' ~---1------t-----+-----t-----+--- t--- -+-----1----I I I I I I I I I I

-10 ----, ----- '-----+-----i------r------l------t-----+----- -----+----- ----o 2 3 5 6 7 8 9 10

TIME (ns)

I I " I I It I I I I I I I I I

10°

10-1

0:: W 3= 10-2 C> 0...

W 10-3 >-

~ 10--4 W 0::

10-a

I I I II I I I II I I I I I I II I I I I I I I III

I L~~ I I I I I I -----~---t-_l_-~..l-LH~-----_!_--_l--~-~-t~_l_w..----11111 111" I 11111:1 I:: 111111 I I I:! : I : 1: I I I .. f I :::: I : I : I I ~~~~-----_+--~--~-~-~..:+~~-----+---~-+-~..:-~1""t ____ + __ ..:--~-+-+..:+:_:_----11111 111111111 111111111 111111111 I I III I I , I , I I " I I I I I II II I I I I , " " 11111 I I I 111111 I I I I I1III I I I 111111

1-~J~~_-----~--~--~-t_+_1+~~----+---t-++_1-~+++----- I --~--+-+-~_1+H-----I I '" I I I I I I I " " I I I , I II I I I I I III I I III I I I I I I III I I I I I I I II I I I I I I I " I I III I I I I I I I II I I I I I " I I I I I I I " II I I III I r I I I I I II I I I I I " It t I I I I " II

1-~ttt------t---~--t-t-t_:+I+-----t---t-+-t_~-~ttt-----+--~- -t-H+H-----I I '" I I I I I I III I I I I I I " I I I I I r II 11"1 I I I 111111 I I I 111111 I I I I 1111 I I "' I I I I I I I " I I I r I I III It' I I I II I I I t I I I I I I I I I I I I I I I I II I I I I I I III - ..... 4 .. ------.... ---1--+-4-....... + ........ -----.. ___ ... _+_ .... -1_ ... +4 .. _____ + __ -1--+-4- ............... ----t I I t I I I I I I I '" I I I I , I II I I I I I I I" I I I I t I I I I I I I II I I I I I t II I I I I I I II 11111 " "'1111 I I I 111111 1 I I I, " 11111 '1'111111 " I 111111 I I I 1111 I 11111 I 1 1111111 I I It 11111 I I I t III I -I"1''t't------,...--.... --1"-"t-t"-1~ ........ -----'t---t"-~__t_ .... -I"1"'t-t-----__t_--_1--1"--t-........ I I I II I I I I I I '" I I I J I I "I I I I I I I I I

10-&

11111 I I I I I111I I I 1 I 11111 I I I I III t 11111 I 1 "1'111 I I I I 11111 I I ,"'11 11111 I I I 1 "ilt I' I 111111 I I , 11111

'-Hii-----+--+-+-i-H+I+-----+---}-+++~+ii-----+--_I--+-i-l-i+I+----

FREQUENCY (MHz)

Figure 1. Typical unipolar UWB pulse and its relative power spectrum by a spark gap generator.

BIOLOGICAL EFFECTS CAUSED BY A SINGLE PULSE

One misconception in studies of the biological effects caused by high peak-power RF pulses is to equate high peak power to high average power because the same power unit (W) and the resultant SAR unit (W/kg) are used in both cases. As indicated previously, high peak-power RF pulses can have a low average SAR if the duty factor is low. Another fundamental issue to address is to decide which of the following independent parameters, SAR or SA, is the most appropriate predictor of biological effects.

Drastic Biological Effects Caused by a Single Pulse of High Energy Deposition

The biological effects ofa single pulse of high energy deposition can be very drastic. An example was the use of high power microwaves for brain enzyme denaturation in the analysis of neurotransmitters. The energy depositions used were as high as 57,000 J/kg in 2.8 s for rats (-20,500 W/kg for 2.8 s) [Lenoxetal. 1977] or 200,000 J/kg in 0.35 to 1.4 s for mice (between 140,000 to 575,000 W/kg) [Schneider et al. 1982]. The brain temperatures reached as high as

209

80°C. These research procedures were designed to achieve enzyme denaturation in as short a

period of time as possible. The susceptibility of rodents to pulsed microwave induced stun and seizure has been

reported. Guy and Chou [1982] noted the threshold for "stun and seizure" (convulsions) in rats by a single 915 MHz pulse. They demonstrated an absorption threshold at 680 J in the brain in less than 0.38 s or equivalent to an energy absorption of28,000 Jlkg. The threshold absorption rate was equal to 73,700 Wlkg for 0.38 s. It should be noted that a very high brain temperature (46°C) was recorded at the threshold for "stun and seizure." The animal began moving again

when brain temperature returned to within 1 ° C of normal. Histological examination revealed

demyelination of neurons one day after exposure and microfocal glial nodules in the brain one month after exposure. Time averaging of this high SAR but short duration threshold is inappropriate because an artificially low SAR can be obtained, i.e., 7.6 Wlkg (680 J /360 s / 0.25 kg).

Modak et al. [1981] studied the effect of a 2.45 GHz microwave pulse of 15 or 25 ms which increased mouse brain temperature by 2 or 4°C, respectively. They used a tuned

waveguide exposure system which delivered 90% of the 7.5 kW transmitted power into the head and 10010 into the brain of the mouse. Immediately after exposure, the spontaneous activity of the exposed mouse decreased but began recovering within 5 minutes. The 25 ms pulse also caused a significant decrease in acetylcholine content of the whole brain. These authors suggested that increased membrane permeability could be the mechanism of the observed change in acetylcholine content.

Microwave Evoked Body Movements

Microwave evoked body movements in mice were demonstrated initially by Wachtel et al. [1988, 1989]. They [Wachtel et aZ. 1988] noted similarities between microwave evoked body movements and classical startle reflexes. Startle is a protective response of mammals to an unexpected, sudden and intense visual, auditory, or tactile stimulus [Fleshier 1965]. Brown et al.

Table 1. Evoked body movements in mice by a single microwave pulse

Peak Brain SAR Pulse Duration Brain SA Incidence

{kWlkg) {ms} {Jlkg) {%}

15.36 50 768 30

15.36 100 1,536 68

15.36 200 3,072 78

9.47 200 1,893 76

2.75 800 2,197 78

0.69 3,200 2,211 29 Data from Brown et al. [1994].

210

[1994] studied this effect in detail in mice exposed locally (head and neck) to 1.25 GHz microwave pulses ranging from 50 ms to 3.2 s. Results of this study are shown in Table 1. The effect appeared to be a complex interaction between peak SAR and SA. For pulse durations less

than 1 s, the incidence of microwave evoked body movements was dependent on SA instead of

peak SAR. Exposure to a single long pulse (3,200 ms) was much less effective in evoking body movements than those caused by shorter pulses. Figure 2 makes it clearly evident that a pulse train (80 pulses per second, 10 J-lS pulse duration, peak brain SAR 1,474 kW/kg, brain SA at 590,

1,180, and 2,360 kJ/kg; and administered between 50 to 200 ms) induced a similar incidence of evoked body movements as those induced by a single pulse. However, the incidence of evoked

body movements decreased if the pulse duration was longer than Is (see open symbols in Fig. 2). Thus, the SA dependency of microwave evoked body movements should be modified as: the incidence of microwave evoked body movements was dependent on average SA if exposure duration was less than Is. In other words, microwave evoked body movements were susceptible

to the influence of temporal summation but possessed a very poor temporal resolution. Part of

this complex interaction between SA and SAR appeared to be related to changes in subcutaneous temperature because the incidence of the evoked body movements was a linear function of the temperature increment. A minimum temperature increment (1.2 0c) for evoking body movements

in mice for exposures shorter than 1 s was noted by extrapolation. The subcutaneous temperature increment which could evoke body movements also appeared to be very narrow. The incidence reached a maximum at a measured temperature increment of 1.66 °C; a total range ofless than 0.5 °C.

~ 99~~~~~~~~~~~~~~~~~~~.-~~~

::J ill A sham O':i • pulse train, 80 pps, 10 J1.s, 1,474 kWjkg peak brain SAR ~ • single pulse, 153.6, 9.47 or 2.75 kW!kg peak Q

w >

~ ::J ::;;0 ::;;0 ::J U

Z ...: VI VI ::J ...: CI

~

90

70

50

30

E 10

w U z w o U z

•

0.0

• ••

• • • •

• 0

0

o 3.2 s long pulse, 0.69 kW/kg peak o 2 s pulse train, 80 pps, 10 J1.s,

147 kW/kg peak

0.5 1.0 1.5 2.0 2.5 3.0 3.5

BRAIN SPECIFIC ABSORPTION (kJ/kg)

4.0

Figure 2. Comparison between short and long pulses. Solid symbols are data from exposures less I s. Open symbols are data from exposures longer than I s [Brown et al. 1994].

Raslear et al. [1992] studied the microwave evoked whole-body movements in LongEvans rats exposed to a single 1 s, 1.25 GHz pulse localized to the head and neck region of the

211

rat in a WR 650 waveguide at graded forward power (40, 80, 120, 160 and 200 W). Dosimetry data was not presented but estimated from Seaman et al. [1992). Peak brain SARs were 800, 1,600, 2,400, 3,200, and 4,000 W/kg and brain SAs were 800, 1,600, 2,400, 3,200, and 4,000 J/kg. They noted that the incidence of evoked body movements increased with power applied to the waveguide exposure system. In addition, a significant increase in the incidence of evoked body movements (15 %) was observed at a subcutaneous temperature increase as low as 0.2 °C

and grew with higher temperature increases (40 % at 0.33 °C, and 80 % at 0.65 0q. It appeared

that evoked body movements required a lower subcutaneous temperature increment in rats than in mice (1.2 to 1.7 °C from 0 to 100 % incidence). Physically, rats have a much larger head and

neck surface area than that of mice. Spatial summation of subcutaneous temperature increment

may be the key to the species difference in sensitivity to microwave evoked body movements.

No reliable difference in deafened rats and normal rats exposed to similar power levels was observed. Therefore, the findings indicated that an associated microwave acoustic effect was probably not the underlying mechanism for microwave evoked body movements.

A pharmacological characterization of the microwave-evoked whole-body movements in rats was performed by Akyel et al. [1993a). A 1.25 GHz, 120 W pulse for 1 s was delivered to the head and neck region of the rat as in the previous experiment [Raslear et al. 1992). This pulse was to evoke a 50 % movement incidence in the vehicle injected animals. Eleven drugs were used: pilocarpine, atropin, clonidine, desipramine, p-choloramphetamine, buspirone, morphine, naloxone, d-amphetamine, pimoxine and diazepam. The movement incidence rates (50%) occurred as expected, and only atropin and clonidine injected rats failed to display microwaveevoked whole-body movement. It was concluded that a microwave-evoked whole-body movement was quite different from the neuropharmacology of a classical startle response since atropin should produce a slight increase in the startle response while clonidine depresses it [Davis 1980).

Startle Modification

Startle can be modified by a pre-pulse, a preceding brief sensory but not startle-evoking stimulus, that is acoustic, tactile or photic in nature. The ability of a microwave pulse to serve as a startle modifier was studied by Seaman et al. [1992, 1993, 1994). The identical 1.25 GHz pulsed exposure system used by Raslear et al. [1992] provided the pulse. Two different microwave pre-pulses, approximately of 1 fls pulse duration, 15.0-30.0 kW /kg (16.0-44.2 mJ/kg) and 35.0-86.0 kW/kg (66.6-141.8 mJ/kg) were delivered 201, 101,51,3 and 1 ms before and 1 ms after the startle eliciting acoustic stimulus (94 dB SPL, sound power level), or a 7.82 flS microwave pre-pulse at 55.9-113.3 kW/kg (525.0-1,055.7 mJ/kg) at 157, 107,57, and 7 ms before and 43 ms after the eliciting startle air burst (tactile stimulus). The lower intensity

microwave pre-pulse (16.0 to 44.2 mJ/kg) could not modify the acoustic startle amplitude or latency. However, acoustic startle amplitude was reduced by the higher microwave pre-pulses (66.6-141.8 mJ/kg) at 51,101 and 201 ms, and enhanced by higher microwave pre-pulse at 1 ms lead time. Acoustic startle latency was also delayed by the higher intensity microwave pre-pulses at 51 and 201 ms lead times. The effect of the highest microwave pre-pulse (525.0-1,055.7 mJ/kg) was not as clear cut in modification of the tactile startle amplitude because the amplitude reduction was observed at a 57 ms lead time only. On the other hand, the latency of the tactile response was delayed when microwave pre-pulses were administered at 7, 57, 107, and 157 ms lead times. It is apparent that the heating potential of these microwave pre-pulses was negligible

212

«10- 3 0C). Nevertheless, they were adequate to elicit modification of ongoing reflexes with a threshold SA higher than the threshold SA that can induce microwave hearing.

Thermal Sensation

Thermal sensation and thermal pain are other biological responses which can be used to elucidate the nature of biological effects of high power RF. Excellent reviews were made by Stevens [1983] on comparative warmth sensation elicited by infrared and microwave radiation and by Michaelson and Lin [1987a] on thermal sensation and pain elicited by microwave exposure. In contrast to infrared illumination which has a short onset and offset latency (abrupt onset and extinction of the sensation), microwave illumination induced a warm sensation that appeared to have a relative long onset latency (3.5 to 6 s) and long offset time (15 s or more), described as a kind of "afterglow" [Justesen et al. 1982]. The former reviews report that above the threshold, intensity ofwannth sensation increases with power density of either radiations. The three most important variables governing the threshold power density are skin location, size of exposure area and duration of exposure. Each variable contributes differently to the threshold power density depending on the level of warmth perception. At a barely perceptible level, threshold power density is inversely proportional to the area of exposure (spatial summation). The area size of the spatial summation may be as large as 60 cm2 or more. It also depends on the location of skin exposure. The forehead is the most sensitive of the areas tested, followed by locations on the torso and extremities. The difference in skin location sensitivity becomes less with increasing levels of warmth sensation. Sensitivity (threshold power density) of warmth sensation is also inversely proportional to the duration of exposure (temporal summation), i.e., the longer the exposure duration, the lower the threshold. At a barely perceptible level, the threshold power density becomes asymptotic when the exposure duration is longer than a critical value. The critical period for an asymptotic threshold appears to be 1 s for infrared radiation [Stevens 1983] and 3 s for 3 GHz microwave radiation [Eijkman and Vendrik 1961]. A summary of microwave-induced wannth sensation thresholds is included in Table 2. In addition to location, area, duration, frequency, and power density, the penetration depth is also an important variable in determining the threshold [Blick et al. 1997]. The threshold can vary more than an order of magnitude (14 fold difference) as shown in Table 2.

The salient feature of human wannth sensation is that the cutaneous temperature increase at the threshold is less than 0.1 °C. On the other hand thermal pain is elicited by a high cutaneous

temperature (45-46 °C) [Michaelson and Lin 1987a]. Irrespective of type of radiation, either

infrared or microwave, human warmth sensation has a relatively low threshold for the rate of temperature change (0.0005 to 0.002 °C/s) across the cutaneous thermoreceptor located

approximately 150 to 200 Jim below the skin surface. The rate of temperature change in human warmth sensation (0.0005 to 0.002 °C/s, 4 fold difference) for 1 to 10 s exposure time

corresponds to 1.7 and 7.0 W/kg average SAR if a specific heat, 0.83 kcal per kg per °c, is assumed for the skin.

Auditory Effect (Microwave Hearing)

Frey [1961] was the first to study systematically the human auditory response to pulsed microwaves. Guy et al [1975] concluded that one of the most widely observed and accepted biological effects oflow average power electromagnetic energy is the auditory sensation evoked

213

in man by pulsed microwaves. The effect appears as an audible clicking, buzzing or chirping sensation originating from within and near the back of the head and corresponding in frequency to the recurrence rate of the microwave pulses. This effect has been the subject of several reviews

Table 2. Thresholds for microwave-elicited human cutaneous warmth sensation

Frequency Exposure Skin Location Threshold Increase in Skin (GHz) Duration (s) and Size (cm2) (mW/cm2) Temperature CC)

2.45 10 Forearm, 107.2 26.7 ± 7.3 (6)

2.45 10 Back, 377 63.1 ± 6.7 (15) 0.054*

3.0 Forehead, 37 58.6 0.025

3.0 2 Forehead,37 46.0 0.040

3.0 4 Forehead,37 33.5 0.060

7.5 10 Back, 377 19.5 ± 2.9 (15) 0.049*

10.0 Forehead, 37 21 0.025

10.0 2 Forehead, 37 16.7 0.040

10.0 4 Forehead, 37 12.6 0.060

10.0 10 Back,377 19.6 ± 2.9 (15) 0.073*

35.0 10 Back, 377 8.8 ± 1.3 (15) 0.078*

94.0 10 Back,377 4.5 ± 0.6 (15) 0.071 * Data shown are mean ± S.E. (no. of subjects) from Hendler [1968]. Hendler et al. [1963]. Justesen et al. [1982]. and Blick et at. [1997]. "*,, indicates data calculated by Riu et al. [1997].

[Chou et al. 1982a, Lin 1978, 1980, 1990, NCRP 1986]. Foster and Finch [1974] showed both theoretically and experimentally in a physiological solution that microwave pulses could produce significant acoustic energy by thermal expansion from 5 x 10-6 0 C temperature rise in the solution

from absorption of microwave pulses. It is generally accepted that the pulsed microwaveinduced audible sound is generated by a thermoelastic expansion of cranial tissue [Lin 1976, 1976a, 1976b, 1976c, 1977, Lin et al. 1975] that launches an acoustic wave which is detected by hair cells in the cochlea. Essentially identical electrical activities elicited by microwave and acoustic pulses were recorded in the auditory pathway of cats and guinea pigs at the primary auditory cortex [Taylor and Ashleman 1974, Chou et al. 1976], the medial geniculate nucleus [Taylor and Ashleman 1974, Guy et al. 1975, Lin et al. 1979, 1982], the inferior colliculus

nucleus [Cain and Rissman 1978, Lin et al. 1979], the lateral lemniscus nucleus [Lin et al. 1979, 1982], the superior olivary nucleus [Lin et al. 1979, 1982] and the eighth cranial nerve [Taylor and Ashleman 1974]. In addition, brainstem potential and cochlear microphonics evoked by microwave pulses were comparable to those evoked by acoustic pulses [Lin et al. 1979, 1982, Chou et al. 1975, 1976].

Taylor and Ashleman [1974] demonstrated the dependency of the microwave-induced

214

auditory effect on cochlea in cats by measuring the electrical responses in the eighth cranial nerve, the medial geniculate nucleus, and the primary auditory cortex to both acoustic and 2.45

GHz microwave pulses. They concluded that evoked electrical responses were similar between

acoustic and microwave pulses. Furthermore, cochlear destruction, i.e., perforation of round window and aspiration of perilymph, resulted in abolishment of evoked electrical potential from acoustic and microwave pulses. Chou et al. [1975] recorded the sonic oscillations at 50 kHz from the round window of guinea pigs during pulsed 918 MHz irradiation. The oscillations promptly followed the onset of pulsed radiation, preceded the nerve response and disappeared after death. It is therefore reasonable to conclude that the pulsed microwave induced auditory effect is a cochlear response to acoustic signals that are generated in the head by pulsed microwaves.

For pulsed microwaves with durations shorter than 30 J.ls, the microwave-induced auditory threshold was independent of temporal peak power density (or temporal peak specific absorption rate) and was entirely dependent on the energy density of the microwave pulse or specific absorption (SA) per pulse [Chou et al. 1982a, Lin 1978, 1980, 1990, NCRP 1986]. The

auditory threshold SA per pulse has been determined in human volunteers (16 mJ/kg), cat (10 to 12 mJ/kg) [Guy et al. 1975], cat (4 mJ/kg) [Lebovitz and Seaman 1977] and rats (0.9 to 1.8 mJ/kg whole-body averaged SA) [Chou et al. 1982a]. The heating potential of these pulses at these auditory thresholds was on the order of 10-6 °C. It is the most sensitive biological effect

induced by microwave radiation. For microwaves with pulse durations longer than 50 J.ls, the auditory threshold is dependent on peak SAR. Microwave-induced pressure waves, their fundamental frequency, and propagation have been characterized in mammalian brain and models with a hydrophone [Olsen and Lin 1981, 1983, Lin et al. 1988]. A near identical result between prediction by the thermoelastic theory and hydrophone data further strengthens this explanation of the basic mechanism of microwave-induced auditory sensation. These studies also pointed out that a pressure wave could be created by a single microwave pulse within the cranial structure.

One of the consequences of the microwave acoustic effect could be annoyance. Annoyance caused by pulsed microwaves was not explicitly demonstrated in humans or animals. Indirect evidence of avoidance has been presented by some investigators [Frey et al. 1975, Frey and Field 1975, Johnson et al. 1976, Hjeresen et al. 1978]. However, the avoidance resulted from exposure to multiple pulses. Critical pulse repetition rate and duration of exposure needed to elicit an avoidance behavior have not been explored.

Lin [1989] proposed that this thermoelastic induced pressure wave could be one of the important mechanisms responsible for pulse modulated RF interactions with biological systems. He has calculated the peak pressure and displacement in various spherical head models irradiated with 10 J.lS rectangular microwave pulses (Table 3). The isolated rat lens was found to displace

by 10 nm when irradiated with a 10 J.ls, 300 J/m 2 microwave pulse [Brown and Wyeth 1983]. Substantial displacement and pressure (10 nm, 170 N/m2) could develop within the cranial structure if the peak power density reaches 5 MW/m2. Considerable damage to cell membrane and cytoplasm could occur by mechanical stress. Some supporting evidence (in vitro lenticular damage by multiple pulses) has been provided by a group of investigators [Stewart-DeHaan et al. 1983; 1985; Creighton et al. 1987].

Miscellaneous Effects Caused by a Single Electromagnetic Pulse

Hirsh et al. [1968] exposed trained rats to an electromagnetic pulse (EMP) and studied their ability to run a maze during and after EMP exposure. The EMP pulse used was 600 kV/m,

215

3 ns duration. "Startle" was noted to be associated with the EMP pulse and the ability of animals to run the maze was compromised. Recovery began 10 minutes after exposure and was nearly complete in 30 minutes after exposure. It was concluded that the animal's decision making strategy had been disturbed in some way, because there frequently was a hesitation at each point in the maze structure where a decision was required as to whether a left or right turn should be taken. The observation was interpreted as evidence of a slowing of mental processes and psychomotor reactions. Although Hirsh et al. [1968] indicated that the EMP pulses were not accompanied by any auditory or visual phenomena which could act as stimuli, one cannot dismiss that the animals may have been disturbed by the auditory stimulus associated with the EMP pulse. Akyel and Raslear [1993] observed characteristic ear ''twitches'' in rats associated with a 70 kVlm EMP pulse. Because ear "twitches" developed regardless of the animal's position in the EMP field, it was suggested that this ear response was caused by the very brief, high pitch sound generated by the discharge of capacitors in the power supply. Sounds produced by capacitor discharge could be heard by investigators in the shielded room which housed the EMP simulator.

Table 3. Peak pressure and displacement in spherical head model irradiated with 10 ~s rectangular microwave pulses at a peak absorption rate of 1 kWIkg

Sphere Microwave Incident Radius Frequency Pressure Displacement Power {mm} (MHZ) SEecies (N/m~ {to" nm} (W/m2}

20 2450 guinea pig 0.408 2.16 4,450

30 2450 cat, monkey 0.369 1.51 5,890

50 918 human-infant 0.961 9.34 12,820

70 918 human-adult 0.682 3.97 21,830 Data adapted from Lin [1989].

COMPARISON OF BIOLOGICAL EFFECTS CAUSED BY MULTIPLE IDGH PEAK POWER RF PULSES AND CW RF OF EQUIVALENT AVERAGE ABSORPTION

Because of the availability of dedicated and specifically designed RF exposure facilities for studying biological effects ofRF radiation, a majority of studies utilized either CW or pulse modulated RF. Few exposure facilities provide the flexibility to change pulse characteristics such as peak power, pulse duration and pulse repetition rate and quantitatively examine associated biological changes. In contrast to several proposed physical mechanisms in explaining biological effects of sinusoidal modulated [postow and Swicord 1996] or of carrierless RF, a general lack of physical mechanisms was proposed to explain the difference in biological effects caused by CW and pulsed RF. The exception to the above statement is a proposed thermoelastic or thermomechanical effect associated with pulse modulation. Neshev and Kirilova [1996] described a theoretical model which indicated pulse-modulated microwaves could potentially influence the conformational osciIlations of enzymes in living organisms and produce effects at extremely low power levels. This hypothesis is still to be proven. Considering equipment and theoretical

216

difficulties, this section evaluates only studies designed to compare the biological effects caused by CW and pulsed RF of equal averaged SAR. However, some studies of historical importance will also be included even if a CW counterpart was not studied.

Behavioral Effects

A behavior study [deLorge 1984] involving rhesus monkeys exposed at near resonant (225 MHz, CW) and supra-resonant frequencies (1.3 GHz and 5.8 GHz) showed that the performance of monkeys trained to press levers in an observing-response paradigm for food was impaired at different threshold intensities for different frequencies. The time averaged threshold intensities were 81 W/m2 (whole-body average SAR= 3.24 Wlkg) for 225 MHz, 570 W/m2 (7.41 Wlkg) for 1.3 GHz and 1400 W/m 2 (4.3 Wlkg) for 5.8 GHz. These threshold intensities were associated with reliable increases in colonic temperature in the range of 1 °C above the sham exposure level. The peak intensities were 81 W/m2 (CW) for the 225 MHz, 518 kW/rrr for the 1.3 GHz (370 pps, 3 I1s), and 1.06 MW/m2 for the 5.8 GHz (662 pps, 2 I1s).

Thomas et al. [1975] compared 30 minute exposure of 50 to 200 W/m2 CW and pulsed 2.86 GHz microwaves on post-exposure performance of rats trained under a multiple fixed-ratio (FR), differential reinforcement oflow rate (DRL) operant schedule. The pulses used were 1 J.l,S

pulse duration, 500 pps and a duty factor of2.5 x 10-4. Instead of absolute response rate, change in behavioral performance was expressed as percentage relative to the performance rate during the preceding sham exposure within each individual rat. They found that the dose-response relationships of behavior endpoints were biphasic or multiphasic in nature. They noted that 200 W/m2 pulsed microwave (800 kW/m2 peak) significantly increased response rate, approximately by 40010 during the DRL portion of the behavior. In addition, a marked increase in the proportion of shorter inter-response times was noted during the DRL portion of behavior when rats were exposed to 100 and 200 W/m2 (400 and 800 kW/rd peak). In conjunction with the DRL changes, FR response rates were lower than those observed after sham exposure irrespective of types of exposures. The largest effect (10% of the control) was found for the 200 W/m2 pulsed exposure. Differences between CW and pulsed microwaves on FR performance were not apparent at other doses. Thomas et ai. [1975] consider the most robust effect was the increased number of time-out responses after exposures. While the number of time-out responses of 50 W/m2 pulsed microwave was within control range, increased time-out responding was increased after the 50 W/m2 CW exposure sessions. Increased time-out responding was found after 100 and 150 W/m2 CW and pulsed exposures and the effects of pulsed exposures was larger in both cases. On the other hand, time-out responding was lower than control in pulsed 200 W/m2• However, time-out responding during CW 200 W/m2 exposure was not evaluated. Because ofbiphasic and multi-phasic response curves, competing events with different dose-response characteristics could be assumed. However, the value of this report was somewhat diminished because the variation of each data point was not included in this report.

Thomas et aI. [1982] trained rats under a differential-reinforcement oflow rate paradigm and compared the efficacy of the CW and pulsed 2.8 GHz microwaves on the number of appropriate responses between 8-12 s of inter-response time (IRT) during each behavior session. A fixed pulse duration (2 J.l,s), pulse repetition rate (500 pps) and duty factor (10.3) were used for pulsed microwaves. Thirty minute exposures of sham, CWand pulsed microwaves at 10, 50, 100, and 150 W/m2 were used. For pulsed microwaves, the corresponding peak power densities were 10, 50, 100, and 150 kW/m2• Rats were exposed at weekly intervals, 2 to 3 times to a mixed

217

schedule of sham, CW and pulsed conditions. Within individual comparisons were made. They found the rate of emission of appropriately timed responses declined after exposure to pulsed microwaves at 100 and 150 W/m2, whereas similar exposure to the CW microwaves did not produce consistent effects. They further noted that the overall responses, regardless of correct or incorrect responses, were not reduced as a result of microwave exposure. It was concluded that the rat's ability to respond was not impaired, but rather that its ability to discriminate the appropriate IRT interval was disrupted.

Lebovitz [1983] compared the effects ofCW and pulsed 1.3 GHz microwaves at two different levels (5.9 W/kg for CW, 6.7 W/kg for pulsed, and 3.6 W/kg for both CW and pulsed) on the performance of rats trained on a multi-component (fixed-ratio, timeout, FR25" TOIO) operant task during exposure. The pulse characteristics were 1 IJ..S pulse duration, 600 pps and

6 x 10-4 duty cycle. A circularly polarized waveguide exposure system was used. Duration of exposure was 3 hours. The results indicated that CW (5.9 W/kg) and pulsed (6.7 W/kg) microwaves were equally effective in reducing response rates during both the fixed-ratio and the timeout component of the operant task. At 3.6 W/kg the mean rates of fixed-ratio responding were unchanged, whereas the rates of responding during timeout were reduced significantly. Again, CW and pulsed microwaves yielded essentially equivalent results. The conclusion from this report was derived from the actual behavioral data from 14 to 15 animals in each treatment group. The author further commented that sensitivity of the operant behavior to microwave suppression was inversely related to the robustness of the operant behavior. Behavior on the FR schedule was relatively unaffected while behavior during the TO portion seemed to be more susceptable to microwave exposure.

D'Andrea et al. [1994] compared the effects of two microwave pulses on the behavioral performance of monkeys. The behavioral task was a complex schedule composed of a variable interval schedule (VI 25 s) on one lever and a color discrimination task on the other lever for food. The pulsed microwaves were 5.62-GHz operated at 100 pps generated by a radar unit (2.8 IJ..S pulse duration, 2.8 x 10-4 duty factor) and by the radar unit whose output was amplified by a Stanford Linear Energy Doubler (SLED) (50 ns pulse duration, 5 x 10-7 duty factor) which

enhanced peak power by a factor of nine by adding a high power pulse to the radar pulse. Sham exposure and three whole-body average SARs (2, 4 and 6 W/kg) were used. To achieve these doses, peak power densities were 0.56, 1.28 and 2.77 kW/m2 for radar pulses and 5.18, 12.70 and 25.2 kW/m2 for the SLED pulses. The duration of exposure was 20 minutes. Compared to sham exposures, significant alterations of lever responding, reaction time, and earned food pellets occurred during microwave exposure at 4 and 6 W/kg but not at 2 WIkg. There were no differences between radar or SLED. These results complimented the earlier finding in de Lorge's [1984] study of the behavioral threshold in monkeys exposed to 5.8 GHz pulsed microwaves. Furthermore, it was concluded that high peak power microwaves did not possess a unique hazard at peak field intensity near the 100 kVlm (-26.5 MW/m2). The incident pulse energy of these

pulses (1.56 to 7.76 mJ/m~ was higher than the threshold for microwave hearing (20-400 ¢/m2)

[Lin 1980, 1990, Chou et al. 1982a], hence, an acoustic effect could be expected. Because of equal pulse energies between radar and SLED pulses, differences in microwave acoustic effects do not seem to have occurred.

That microwave pulses can serve as a discrimination cue in behavioral situations is supported by works of several investigators [Frey et al. 1975, Frey and Field 1975, Johnson et al. 1976, HJeresen et al. 1978]. Johnson et al. [1976] trained rats to nose poke for food pellets on an FRS schedule in the presence of an 7.5 kHz acoustic cue (10 pps, 3 ~s). When a 918 MHz

218

microwave signal (150 W/m2 average, 10 pps, 10 ~s, 1.5 MW/m2 peak power density) was used to substitute for the acoustic cue or administered during an extinction period (no acoustic cue was given), the response rate of the animals was similar to that during the period when the acoustic cue was present. Frey et al. [1975] used a shuttle box with shielded and unshielded sides and studied the amount of time that rats spent in each side of the shuttle box. A 1.2 GHz microwave was presented as CW (24 W/m2, 2.2 Wlkg whole-body averaged SAR) or pulsed (10 W/m 2

average, 1 Wlkg whole-body averaged SAR, 21 W/m2 peak, 1,000 pps, 0.5 ms pulse duration).

During the last 2 of 4 successive daily 30 min exposures, rats exposed to pulse modulated microwaves spent less time (30 %) in an unshielded side of the shuttle box than the CW exposed rats (64 %) spent in the unshielded side. Frey and Feld [1975] also noted similar preferences (29% in the unshielded side of the shuttle box) in rats exposed to 1.2 GHz pulsed (100 pps, 3 ~s) microwaves at two levels (4 W/m2 average, 1.33 kW/m2 peak; 9 W/m2 average, 3 kW/m2 peak).

Sham exposed rats spent 57 % of their time in the unshielded side of the shuttle box. Hjeresen et al. [1978] used a more sophisticated paradigm to eliminate the original side preference. The microwave pulses used were 2.88 GHz pulsed microwaves (95 W/m2 average, 330 W/m2 peak, 100 pps, 2.3 ~s). They found a similarity between microwave pulses and conventional auditory cues in motivating side preference. These results and the absence of side preference when broadband noise was administered along with microwave pulses, led to the conclusion that an auditory stimulus perceived during pulsed microwave exposure was capable of mediating the preference response.

Ocular Effects A s~ries of experiments was performed using isolated rat lenses exposed in vitro to 918

MHz pulsed microwaves (peak absorption rate= 480,000 Wlkg, 10 J-LS pulse width, and various

repetition rates) for 5 to 6 minutes [Stewart-DeHaan et al. 1983; 1985; Creighton et al. 1987]. A very high perfusion rate (1,600 mllmin) was used to prevent excessive heating «0.65 cC). An

absolute threshold for inducing holes in lens fibers at 231 Wlkg average SAR for a 6 min exposure was estimated in these studies. It was suggested that the lenticular damage could be due to thermoelastic expansion and the result of pressures based on the observation that large globules formed in the exposed lens were not normally seen except at elevated temperatures of 47 and 50 cC in lens in vitro. CW was less effective in inducing lenticular injury than pulsed microwaves.

Birenbaum et al. [1969] compared the effectiveness of the CW and pulsed (0.5 J-LS pulse

duration, and 0.001 duty factor, i.e., 2,000 pps) 5.8 GHz microwaves on the production of lenticular opacity (cataract) in anesthetized rabbits. One eye was exposed while the other served as control. A dielectric loaded waveguide with 1.27 cm diameter aperture was used as a contact applicator. Supra-threshold power densities for cataract formation were selected. Averaged aperture power density was 4.34 to 7.89 kW/m2 (4.34 to 7.89 MW/m2 peak aperture power density) for pulsed microwave and 4.73 to 7.89 kW/m2 for CW microwave. Duration of exposure range from 1.5 to 90 minutes. One hundred rabbits were used for each of the two exposure protocols. The endpoint was cataract formation one month after exposure. The incidence of cataract formation was found to be an inverse relation between aperture power density and duration of exposure. EDso (50% effective dose) of the aperture power density and duration of exposure from each exposure protocol was constructed and compared. It was concluded that a difference in effectiveness of CW and pulsed microwaves in inducing cataract was not found. Acute inflammatory reactions ofthe cornea, conjunctiva, iris, and/or ciliary body were observed in many eyes and probably produced in every exposed rabbit eye. No comparison for these acute

219

ocular inflammatory reactions was made between pulsed and CW exposures. An assessment of injury to rabbit corneas in vivo induced by high peak-power 35 GHz

microwaves (156 kW/m2 peak power density, 21.8 kW/kg peak SAR, 10 to 20~s pulse width, variable repetition rate, 15 min exposure) was performed by Trevithick et al. [1987]. Light microscope and scanning electron microscope were used to evaluate the cornea morphology. They found that cornea damage was dependent on average SAR. The threshold for a single cell destruction was approximately at 33 WIkg. Furthermore, the area of involvement enlarged as the average SAR increased until large areas of many cells were completely destroyed at 550 W/kg. However, a CW counterpart was not performed. Kues et at. [1985] observed corneal endothelial abnormalities in anesthetized cynomolgus monkeys exposed to CW and pulsed 2.45 GHz microwaves. They concluded that pulsed microwaves (100 W/m 2 or 2.6 W/kg average, I kW/m 2

peak, 100 pps, 1O~) were 2 to 3 times more effective than CW (200 to 300 W/m~ in inducing corneal endothelial abnormalities. This conclusion was derived after a variable number of 4 hour exposures, a variable number of examinations, and repetitive use of ketamine and halothane. There were no attempts by other investigators to verify this pulse enhancement effect.

Cardiovascular Effects

Pressman and Levitina [1963] compared the effects of2.4 GHz CW (70-120 W/m2) and pulsed 3 GHz microwaves (30-50 W/m 2 average, 43-71 kW/m2, 1 ~s pulse duration, 700 pps, 7 X 10-4 duty factor) on the heart rate of rabbits. They administered the incident fields from dorsal and ventral surfaces onto 6 different configurations: entire dorsal surface, dorsal back surface, dorsal head surface, entire ventral surface, ventral stomach surface and ventral head surface. A coefficient of chronotropic effect (K) was defined by % of cases showing acceleration (A) and slowing (S) by K = (100 + A) I (100 + S). Positive chronotropic effect (accelerated heart rate) was defined as K > 1, negative chronotropic effect (slowing of the heart rate) as K < 1, and no effect as K = 1. They concluded that dorsal exposure led to a positive chronotropic effect while ventral exposure led to a negative chronotropic effect. For dorsal exposure, the magnitude ofK was higher in pulsed exposed animals than that of CW exposed animals despite a lower averaged incident power density. The magnitude of negative chronotropic K was found to be proportionately greater with larger irradiated surface in the CW ventrally exposed rabbits. An inverse relation between K and irradiated surface was noted in pulsed ventrally exposed rabbits. They interpreted that the negative chronotropic effect was caused by activation of skin receptors and the positive chronotropic effect by brain activation. However, the maximum changes in heart rates were +5 and -8 beats per minutes which are well within the range of spontaneous variation.

Frey and Seifert [1968] showed in the isolated frog heart that a pulsed 1.425 GHz microwave (6 JIW/m2 average, 600 W/m 2 peak, 1,000 pps, 10 JIs) administered at a synchronous period with the R wave in the ECG (220 ms after the P wave) resulted in tachycardia (increased heart rate) or arrhythmia in the isolated frog heart. However, other reports [Liu et at. 1976, Clapman and Cain 1975, Chou et at. 1980] failed to find this pulse synchronizing (cardiac pacing) effect. The pulse characteristics used by Liu et at. [1976] were 1.42 GHz, 320 JIW/m2 average, 3.2 kW/m2 peak, 1 pps, 100 JIs). Clapman and Cain [1975] attempted to replicate Frey and Seifert's pulse and method of study but did not observe cardiac pacing. In addition, the heart rate did not change in studies with a different peak power (55 W/m2), a different carrier frequency (3 GHz) or different pulse durations (2 and 150 JIs). Chou et al. [1980] could not find the pulse synchronizing effect on heart rate in rabbits exposed in vivo to pulsed 2.45 GHz microwaves (10

220

fis pulse, 137 W/m2 peak, pulse duration not specified) triggered by the ECG R wave at various delays (0, 100 and 200 ms) corresponding to R, T and P waves of ECG. Pakhomov et al. [1995]

studied frog auricle contraction rate in an in vitro preparation. The frog auricle was exposed to pulsed microwaves at 915 or 885 MHz for 2 min with a peak specific absorption rate range between 100 and 3,000 W/kg, a pulse duration from 10-6 to 10-2 s and a fixed duty factor (7 x 10-5). Tachycardia did not occur unless the perfusate temperature increased by more than 0.1 ° C.

Apparently, low intensity pulsed microwaves synchronized with the ECG were ineffective in inducing detectable changes in the heart rate.

Hamrick and McRee [1980] assessed the effects of body temperature and pulsed and CW microwaves on the heart rate of embryonic quail. They exposed the embryos to 2.45 GHz CW microwaves for 5-10 min (SARs= 3,6, 15, and 30 W/kg) at incubation temperatures from 35 to 38°C, and to pulsed microwaves (6 kW/kg peak SAR, 10 f-lS pulse duration, 10-50 pps, 0.5 x

10-3 duty factor, and 0.3, 1.5 and 3 W/kg average SAR) at incubation temperatures of35 to 39 °C. There were no significant differences between heart rates of the exposed and control

embryos in any of the groups at any of the temperatures used. They did observe that embryonic heart rate increased approximately 23 beats/min for each 1 ° C rise in incubation temperature

between 36 and 39°C. Birenbaumetal. [1975] evaluated the effects of2.8 GHz CW and pulsed microwaves on

heart rates, respiration rate and subcutaneous temperature in rabbits. Four power densities (200, 400, 600 and 800 W/m~ were used for CW exposures but only 200 W/m2 was used in the pulsed microwave exposure. The pulse characteristics were l.3 f-lS pulse duration, 1,000 pps, l.3 X 10-3

duty factor, and 154 kW/m2 peak power density. Dose dependent increases in heart rate, respiration rate and subcutaneous temperature were noted in rabbits exposed to CWo Between 200 W/m2 CW and pulsed microwaves, no differences in any of three endpoints were found.

Frei et al. [1988] compared the cardiovascular effects of 2.8 GHz pulsed and CW microwave in ketamine anesthetized rats. Power densities in CW exposures were 300, 450, and 600 W/m2 (8.4, 12.6, and 16.8 W/kg whole-body average SAR). Microwave pulses were 2 f-lS

pulse duration, 500 pps, and 10-3 duty factor at average power densities similar to those used in CW exposures and an additional average power density at 750 W/m2 (21 W/kg). Significant increases in heart rate were observed under the pulsed conditions except at 300 W/m2 (8.4 W/kg). The results of this initial study and a follow-up study [Frei et al. 1989a] are compared in Fig. 3. Frei et al. [1989a] re-examined cardiovascular effects with far-field whole-body exposure at 2.8 GHz and whole body average SAR at 14 W/kg for CW and pulsed microwave exposures. Both E- and H-orientations were used, 510 W/m2 was used for both CW and pulsed microwaves (1 MW/m2 peak, 1,000 pps, 0.5 f-lS pulse duration, 0.5 x 10-3 duty factor, SA= 14

mJ/kg) in E-orientation, and 730 W/m2 for both CW and pulsed microwaves in H-orientation (l.46 MW/m2, 1,000 pps, 0.5 f-lS pulse duration, 0.5 x 10-3 duty factor, SA= 14 rnJ/kg). The

cardiovascular endpoints of this study were heart rate and mean arterial blood pressure in the exposed animals when their colonic temperature reached 38.5, 39.0 and 39.5 °C and when

colonic temperature returned to 39.0 °C after exposure. Sham exposure or hyperthermia induced by other modalities was not investigated. Although the time needed to increase colonic temperature from 38.5 to 39.5 °C differed between E- or H-orientation, being shorter in Horientation, both heart rate and blood pressure increased significantly from 38.5 to 39.5 °C (Fig. 3) [Frei et al. 1989a]. The magnitude of changes in heart rate (30 to 41 beats per minute) and blood pressure (9 to 14 mm Hg) were statistically significant and robust. Differences between

221

CW or pulsed exposure in both orientations were not evident. In the earlier report, Frei et al. [1988] showed a significant but lower degree of hyperthermic tachycardia by pulsed microwave exposure when the rectal temperature was increased from 38.5 to 39.5 °C by 12.6, 16.8 and 21 W/kg (Fig. 3). On the other hand, heart rate did not increase significantly even when the rectal temperature was increased from 38.5 to 39.5 °C by 8.4 W/kg pulsed microwave or by 8.4, 12.6 and 16.8 W /kg CW microwaves. The only change of mean arterial blood pressure in this earlier

Data from Frei et al. 1988, 1989a, 2.8 GHz 50 ,----,-----,-----,-----,-----,,-----,----,

E CL

~ 40

30

h: i:5 20 :r:

10

1988

c::=:J CW

e2Zl PW

.. * .. p< 0.05

-10 ~ __ ~ ____ ~ ____ -J ____ -J ____ -J ____ ~~ __ ~

~

0>

20 r----,-----,-----,-----,-----,,----,,----,

:r: E 15

...§. ..... 10 0:: ::::> V) ~ 5 0:: a..

..... <.!)

~ -5 :r: u

19a90

_cw ~PW

*

-10 ~ __ -i ____ ~ ____ -J ____ ~ ____ -J~ __ ~~ __ ~

8.4 12.6 E-14.0 H-14.0 16.8 21.0

AVERAGE SAR (W/KG)

Figure 3. Comparison of cardiovascular effects between CW and pulsed 2.8 GHz microwaves. This figure was ploted from data presented by Frei et al. 1988, 1989a.

report was a decreased blood pressure (hypotension) when the rectal temperature was increased from 38.5 to 39.5 °C by a 16.8 W/kg CWexposure. Differences between pulsed and CW microwaves on heart rate and arterial blood pressure were also not significant [Frei et al. 1988]. Further comparisons between CW and pulsed microwaves on cardiovascular endpoints were made by Frei et al. [1989] in anesthetized rats exposed to 9.3 GHz radiation. Two average power densities (300 and 600 W/m2, SAR= 9.3 and 18.6 W/kg) were used. A fixed duty factor (10.3

from 2 J.lS pulse duration and 500 pps) was used. Thus, peak power densities were 300 and 600 kW/m2. Heart rate and mean arterial blood pressure were again evaluated from 38.5 to 39.5 and back to 39.0 °C at 0.5 °C intervals. A significant increase in heart rate (17 to 23 bpm) was noted when rectal temperature was raised from 38.5 to 39.5 °C. Differences among pulsed and CW microwaves at 300 and 600 W/m2 were not noted. None of these four exposures altered the mean arterial blood pressure significantly. Reconciling these different effects under similar circumstances is difficult.

222

The difference in cardiovascular responses to microwave exposure at different frequencies was interpreted as a result of difference in depth of penetration causing a different degree of skin heating [Frci et al. 1989]. On the other hand, the inconsistencies of the cardiovascular response among microwaves operated at various carrier frequencies could be caused by the instability of the preparation and the central depressant effect of the anesthetic agent used. Depressed responses of the cardiovascular system to microwave induced hyperthermia were clearly shown by the same group of investigators [Frei and Iauchem 1989] in unanesthetized and anesthetized rats exposed to 2.8 GHz ew microwaves. Significant increases in heart rate (-50 bpm) and blood pressure (-16 mm Hg) were observed in the unanesthetized rats while no changes (-1 to 2 bpm and - 1 mm Hg) were noted in the anesthetized rats when their rectal temperatures were increased from 38.5 to 39.5 °e by a similar microwave exposure. In fact, significant increases

in heart rate and blood pressure were noted in unanesthetized rats but not in the anesthetized rats when their rectal temperatures were increased from 38.5 to 39.0 °e, a mere 0.5 °e increase by ew microwave exposure.

Lu et al. [1992] used dorsal head-and-neck exposure (1.25 GHz ew and pulsed) in a waveguide exposure system. Three different whole body average SARs (0, 3.0, and 9.7 W/kg) were used for ew and pulsed exposures. Local brain SARs were 0,9.5,30.4 W/kg, and local neck SARs were 0,34.3 and 110 W/kg. Peak power of the microwave pulses were capable of producing a peak absorption rate at 0.60 MW/kg whole-body average, 1.9 MW/kg brain, and 6.9 MW /kg neck. Two pulse repetition rates and two pulse durations were used to match the low and high ew exposures. They were 0.5 pps, 10 f,LS (duty factor= 5 x 10-5, SA= 6.0 J/kg wholebody average, 19 J/kg brain and 69 J/kg neck), and 16 pps, 1 f,LS (duty factor= 1.6 x 10-5, SA= 0.6 J/kg whole-body average, 1.9 J/kg brain and 6.9 J/kg neck). Endpoints of the study were heart rate, systolic, diastolic and mean arterial blood pressures. Pulse pressure (difference between systolic and diastolic pressures) was also included. Pre-exposure baseline and shamexposure were used for individual and treatment controls. They noted an abnormal decrease in heart rate (bradycardia) while body temperature was increasing during 1.5 to 6 minutes of exposure, and an incomplete recovery in 6 min after exposure. This bradycardia effect was dependent on SAR in terms of number of animals affiicted with the change and the magnitude of change. The pulse pressure also decreased in high SAR groups. At the appearance of bradycardia (2 min into exposure), the increase in colonic temperature could be as little as 0.14 °e while neck temperature increased by 0.59 °e. The difference between potencies ofeW and

pulsed microwaves in inducing these cardiovascular effects was inconclusive due to the large variability in individual animals.

Using an open-end coaxial exposure system, Seaman and DeHaan [1993] compared the effects of 190 s exposures to 2.45 GHz ew, pulsed, and square-wave-modulated microwaves on inter-beat intervals of aggregated cardiac cells from a chicken embryo. The four SARs used in ew microwave exposures were 1.2-2.1, 8.4-12.2, 42.6-48.3 and 72.0-89.9 W/kg. The pulsed microwaves (10.9 f,LS pulse duration, 10,000 pps, 0.109 duty factor) were administered at two SARs only, 1.2-2.1 (peak SAR= 11-19 W/kg) and 8.4-12.2 W/kg (peak SAR= 77-112 W/kg). The gated ew microwaves were modulated at 16 and 0.8-1.7 Hz with a 0.50 duty factor at two higher SARs, 12.0-12.2 and 42.0-43.5 W/kg (24.0-24.4 and 84.0-87 W/kg peak SAR). Results of the study indicated that the inter-beat intervals decreased (i.e., increased heart rate) during ew or gated ew exposures at 42.0 W/kg or higher. The tachycardia effect was consistent with an effect of elevated temperature on heart rate. Alteration in sensitivity to a tachycardiac effect by

223

pulse modulation could not be evaluated because comparable SARs (> 12.2 W/kg) were not evaluated. However, the authors noted a transient decrease in the inter-beat intervals at 8.4-12.2 Wlkg during the 95 s post pulsed exposure period, not the subsequent 95 s sampling period. Increase in the inter-beat intervals (slowing of the heart rate) was noted during CW exposure at 1.2-2.1 Wlkg between sampling time points during the beginning and end of exposure periods, but the change was not significantly different from the pre-exposure baseline. The observations of "low SARs" « 8.4-12.2 Wlkg) on the beating rates were considered by the authors to be inconsistent with a thenna! effect. Because of an unusual large standard error assocated with the transient post-exposure tachycardia of the 8.4-12.2 Wlkg pulsed exposed aggregate and a slower beating rate during the pre-exposure period, this isolated case of post-exposure transient tachycardia could be a chance error.

Wolke et al. [1996] measured the intracellular calcium concentration of isolated cardiac myocytes of the guinea pig exposed to CWand pulsed microwaves used in cellular phones (GSM standard). A host of exposure conditions was studied. In comparison to sham exposed myocytes, decreased intracellular calcium concentration was noted in myocytes exposed to 900 MHz, 50 pps pulses at 30 mW/kg (peak SAR= 59 mWlkg, duty factor= 0.5). The authors concluded that this small difference (decrease from 0.011 ± 0.014 to 0.001 ± 0.009, mean ± S.D.) was not regarded as a relevant effect of the exposure. In comparison to other studies, the duty factors used in this study were very high (0.14-0.80).

Effects on Blood-Brain Barrier Integrity

Frey et al. [1975] evaluated the integrity of the blood-brain barrier by comparing the fluorescein dye concentrations in brain slices in rats after 30 minute exposures to 1.2 GHz microwaves as CW (24 W/m2, 2.2 Wlkg whole-body averaged SAR) or pulsed (10 W/m2

average, 1 Wlkg whole-body averaged SAR, 21 W/m2 peak, 1,000 pps, 0.5 ms pulse duration). The fluorescein dye was found mostly in the lateral and third ventricles of the brain. The dye concentration in the brain slice was higher in both CW or pulse exposed animals than in control animals. Pulsed microwaves induced a similar but more pronounced increase in dye concentration than CW microwaves. Oscar and Hawkins [1977] also compared the efficacies of 20 minute exposures between CW and pulsed 1.3 GHz microwaves on the disruption of the blood-brain barrier. They concluded that brain permeability increased, depending on molecular weight (mol wt) of the tracer, for mannitol (mol wt= 182.2) and inulin (mol wt= 5,000) but not dextran (mol wt= 60,000 to 75,000) in rats exposed to CWand pulsed microwaves at average power densities lower than 30 W/m2• Increased permeability was observed immediately and 4 hours after exposure, but not 24 hours after exposure. Dose response curves were presented for CW, and two different pulses on brain uptake of mannitol at average power densities lower than 30 W/m2•

Biphasic response curves were noted for CW and one of the pulsed microwaves (1,000 pps, 0.5 J.LS pulse duration, 0.0005 duty factor) and the pulsed microwave was less effective in increasing the brain uptake of mannitol than the CWo On the other hand, wider pulses with a lower repetition rate (5 pps, 10 J-lS pulse duration, 5 x 10-5 duty factor) were much more effective on increasing the brain mannitol uptake than CW of equivalent average power densities. The sensitivity and time-course of blood-brain barrier interruption caused by microwaves was also noted by Albert [1979]. Albert [1979] used Chinese hamster and 2 hour exposures to 100 W/m2, 2.8 GHz CW microwave. Gross and ultrastructure observations of horseradish peroxide in the brain were used to evaluate the blood-brain barrier integrity. Immediately after exposure, leaky

224

or disruption of the blood-brain barrier was indicated by the dark-brown stain in brain slices and the electron dense horseradish peroxidase product surrounding the affected capillaries in the brain. A different time-course was noted, i.e., impermeability of blood-brain barrier to horseradish peroxide was re-established in two hours after exposure. In addition, the brain temperature increase was less than 0.4 DC when blood-brain barrier disruption occurred.

The blood-brain barrier effect of microwave exposure was not confirmed by Preston et al. [1979] who used the same Oldendorfprocedure [Oldendorf 1970] as Oscar and Hawkins [1977] and did not find changes in brain uptake index of mannitol in medulla, cerebellum, diencephalon, and cortex in rats exposed to 2.45 GHz CW microwaves at 1 to 300 W/m 2 for 30 minutes. Merritt etal. [1978] duplicated some of the pulsed parameters (1,000 pps, 0.5 ms pulse duration and 30 minute exposure) but extended the peak powers from 20 to 150,250, and 750 W/m2 which resulted in average power densities of 10,75, 125, and 375 W/m 2• They noted that fluorescein concentration in various areas of the brain increased with power density. However, none of pulsed exposures increased brain fluorescein concentration statistically. For positive control, hypertonic urea and ambient heating (43 DC for 30 min. resulted in more than a 4 DC increase in brain temperature) were found to be effective in disrupting the blood-brain barrier integrity. An Oldendorf [1970] double isotope procedure for studying blood-brain barrier was also used by Merritt et al. [1978]. Four different 1.3 GHz microwave pulses were used. Their average power densities were 20 W/m 2 (2 kW/m2, 1,000 pps, 10 Jl.S pulse duration), 200 W/m 2

(20 kW/m2, 1,000 pps, 10 Jl.S pulse duration), 3 W/m 2 (6 kW/m2 peak, 50 pps, 10 Jl.S pulse duration) and 15 W/m 2 (30 kW/m2, 50 pps, 10 Jl.S pulse width). The comparable CW counterparts were not used. Neverthless, CW exposures were 1, 10, 100 and 500 W/m 2• Sham exposed animals and 30 minute ambient heated (45 DC) animals injected with serotonin (50 mg/kg, i.p.) and animals injected with hypertonic urea were used for control and positive control. Brain uptake index of mannitol was used as a measurement of the blood-brain leakage. Again, no effect on blood-brain integrity was noted in rats after 30 minute exposures to pulsed microwaves. The exception was the increased manitol uptake index in ambient heated serotonin injected and urea injected animals. Fluorescein leakage into the brain parachyma was also used by Merritt et al. [1978] for evaluation of the blood-brain barrier integrity in rats exposed to 1.2 GHz pulsed microwaves at 0, 20, 150, 250 and 750 W/m 2 average power densities at fixed duty factor (0.5) from 1,000 pps and 0.5 ms pulse duration. Thus, the peak power densities were 0, 40, 300, 500 and 1,500 W/m 2. The CW counterparts were not included. A positive control was a 30 minute ambient heating at 45 DC. With the exception of positive controls, none of exposed groups showed increased fluorescence in brain slices. Gruenau et al. [1982] also reported no significant change on the penetration of14C-sucrose into the brain of rats after a 30 minute exposure to 2.8 GHz microwaves either as pulsed radiation (2 Jl.S pulse duration, 500 pps, 10-3 duty factor) of various intensities (10 to 150 W/m~ or as CW radiation of various intensities (100 to 400 W/m2).

In contrast to these negative results on the disruption of the blood brain barrier, several studies report significant increases in brain temperature as a result of appropriate microwave exposure with consistent blood-brain barrier disruption and graded severity of the disruption. Lin and Lin [1980, 1982], Goldman et al. [1984] and Williams et al. [l984a] all have demonstrated that increased dye or tracer leakage into the brain tissue would not occur unless microwave exposure caused at least a 4 to 5 DC increase in brain temperature. At this brain temperature (42 DC), the colonic temperature of the whole-body exposed rat was usually maintained at 41.5 DC which could result in reduced renal excretion of tracers [Kanter 1960], therefore, blood

225

concentration of the tracer could rise which increased brain uptake of the tracer. Oscar et al. [1981] noted increased blood flow in all 17 brain regions by 39% to more than 100% in rats after 60 minute exposures to a 2.8 GHz pulsed microwaves (2 J-LS pulse duration, 100 pps, 150 W/m2

average). Therefore, brain uptake of tracers could be enhanced by the increased brain perfusion rate. The tracer concentration in the brain parachyma caused by increased brain blood flow can

mimic changes in endpoints similar to those caused by a low level increase in blood-brain barrier permeability.

On the other hand, localized brain exposure was used in experiments performed by Lin and Lin [1980, 1982] and Goldman et al. [1984] and less than a 1 °C change in rectal temperature

was noted. Although an indirect renal effect on blood tracer concentration was less important in the experiments using localized brain exposure, the hyperthermic enhancement of brain perfusion could not be entirely dismissed. The dependency of blood-brain barrier disruption on brain hyperthermia was further demonstrated by Neilly and Lin [1986]. Graded doses of ethanol (0, O. 1, 0.3, 0.5, and 0.7 g%g body weight) were administered to reduce the magnitude of brain temperature increase in rats caused by a constant local brain exposure (3.15 GHz, CW, 30 kW/m2, 15 minutes). The graded reduction in severity of blood-brain barrier disruption was found to be correlated with a reduction in brain hyperthermia by graded doses of ethanol. Sutton and Carroll [1979] demonstrated that the mortality and severity of blood-brain barrier disruption was less if the lower body of the animal was precooled to 30°C while the brain temperature was

maintained at a threshold hyperthermic level of 40 to 45°C by microwave exposure. Additional studies on effects of microwaves on the permeability of horseradish peroxidase,

and sucrose through the blood-brain barrier were also performed by Williams et al. [Williams et al. 1984b, 1984c, 1984d]. Microwave radiation and blood-brain barrier function has been studied extensively in the past. An additional review on the topic can be found in an article by Lai [1994]. It was concluded that later studies were not supportive ofthe earlier findings, the disruption of blood-brain barrier by low SAR microwaves, by Frey et al. [1975] and Oscar and Hawkins

[1977]. In fact, it was concluded that suppression of blood-brain barrier permeability occurs as a result of microwave exposure, and that this effect is mediated by temperature-dependent changes in endothelial cell function, and not by quantities unique to microwave energy [Williams et al. 1984e].

Effects on Nervous System

Baranski [1972a] compared the effectiveness of3 GHz CW and pulsed microwaves (400 pps) on brain histology and histochemistry in guinea pigs. Six different microwave treatments were used. They were 35 W 1m2 in CW and pulsed for 3 hours daily for 3 months, one 35 W/m2

in CW and pulsed for 3 hours and one 250 W/m2 in CW and pulsed for 3 hours. No other pulse characteristics were included. He concluded that chronic repeated microwave exposure led to a morphologic lesion indicative of metabolic disturbances in myelin sheaths and glial cells as expressed by the appearance of peculiar metachromatic spherical bodies. Glial cell proliferation, decreased acetylcholinesterase activity and decreased acid dehydrogenase activity were also noted. These effects were more profound after exposure to pulsed microwaves than after exposure to CW microwaves. Due to lack of information, the dependence of pulse enhancement on peak SAR or SA could not be evaluated.

Seaman and Wachtel [1978] compared the effectiveness of 1.5 and 2.45 GHz pulsed and CW microwaves on the firing rate of the Aplysia pacemakers. They found that a few W /kg

226

exposure for 2 to 3 minutes could change the firing rate of Aplysia pacemakers. While a difference between pulsed and CW microwaves on the slow response was not apparent, pulsed microwaves could induce a rapid response more readily than did CW radiation at the same SAR. An earlier report [Watchel et al. 1975] by the same group of investigators indicated that they were unable to detect any significant differences in the effects of pulsed (10 J)-S pulse duration, at

1,000 and 5,000 pps) versus CW microwaves on Aplysia pacemakers exposed at 10 W/kg. Details of pulse characteristics in both reports were not provided by these authors other than that a 0.5 to 10 J)-S pulse duration and 1,000 to 5,000 pps were used.

Chou and Guy [1978] compared the effect of pulsed and CW 2.45 GHz microwaves on compound action potentials of isolated frog sciatic nerves, cat saphenous nerves, rabbit vagus nerves and superior cervical ganglia, as well as contractility of rat diaphragm muscles in a waveguide exposure system with circulating Ringer's solution to maintain a constant bath temperature during exposure. Graded peak SARs (0.3,3,30 and 220 kW/kg) were used. A constant duty factor (10-3) was used by combining two pulse repetition rates (1,000 pps and 100 pps) and two pulse durations (1 J)-S and 10 J)-s) to obtain average SARs at 0.3, 3, 30, and 220

W /kg. They concluded that there were no significant changes in characteristics of nerves (compound action potentials, CAP) and muscles exposed to CWat SARs of 0.3-1500 W/kg and pulsed peak SARs of 0.3-220 kW/kg (average SAR of 0.3-220 W/kg). No direct stimulation of nerves by either CW or pulsed microwaves was observed. At the higher specific absorption rates, the CAP showed a slight increase in conduction velocity but no change in amplitude. This result was consistent with a 1 ° C increase in bathing solution temperature wherein the results were

replicated by increasing the bathing solution temperature of an unexposed preparation. In a series of two reports, McRee and Wachtel [1980, 1982] compared the effectiveness

of pulsed and CW 2.45 GHz microwaves on altering the "vitality" of isolated frog sciatic nerves stimulated with twin electric pulses separated by a 5 ms interval at 50 pulse pairs per second. "Vitality" was defined as the half-decay time ratio of the CAPs between exposed and control nerves. Both CAPs were examined. Microwave pulses used were 10 W/kg average SAR, 20 kW/kg peak SAR, 10 J)-S pulse duration, 50 pps and 0.5 x 10-3 duty factor. The microwave pulses

were delivered in three different phases of nerve firing: (1) asynchronous, wherein the microwave pulse was delivered at varying times in the nerve firing cycle; (2) synchronous, with the peak of the action potential; (3) synchronous, with the quiescent period between nerve firings. A 20 to 30 minute exposure was used. In all pulsed microwave exposures, the "vitality" decreased in comparison to the unexposed mate (control). However, the magnitude of this effect was essentially the same in all three cases and was also comparable with the effect seen using CW microwaves of equivalent SAR.

Arber and Lin [1985a] noted that snail neurons exposed to CW microwaves (12.9 W/kg) at constant temperatures (8 and 21 0c) resulted in inhibited spontaneous activity and reduced

input resistance. On the other hand, modulated (amplitude modulated at a radom-noise frequency, 2-20 kHz) microwaves at 6.8 and 14.4 W/kg predominately caused excitory responses and increased membrane resistance. It was concluded that the microwave effect on snail neurons was related to the release of intracellular calcium ions (Ca++) [Arber and Lin 1985a]. The conclusion was based on the observation that intracellular injection of EDT A, a chelating agent, completely eliminated the microwave responses in snail neurons [Arber 1981] and the manipulation of extracellular Ca++ concentration had not influenced the fall of membrane resistance induced by microwaves [Arber and Lin 1985b]. Ginsburg et al. [1992] re-evaluated the principle that weak microwave fields could perturb a nonequilibrium physiochemical process

227

in an excitable tissue at a constant system temperature. They could not replicate the original findings of Arber and Lin [1985a] and concluded that microwave irradiation might enhance metabolic rundown and/or loss of ion channel function which occurred over time. In addition, a specific target for microwave interaction with neurons was not suggested by them [Ginsburg et al. 1992].

Chou et al. [1982b] compared the effects CW and pulsed microwaves on the electroencephalogram and visual and auditory evoked potentials. The microwaves were 2.45 GHz at 15 W/m2 average, 2 hours per day for 3 months. The pulse characteristics were 10 J.ls

pulse duration, 100 pps and 10-3 duty factor. Six rabbits (3 males and 3 females) were used in sham-, CW- and pulsed-exposure groups. Frequency spectrum and amplitude of sensory-motor cortex and occipital cortex were evaluated over the 3 month exposure period. Due to considerable variations from animal to animal and from recording session to recording session, comparison was difficult. Other than a general trend of decreased amplitude during the later part of the experiment, no difference was found in the EEG endpoints. Similar variations also existed in visual and auditory evoked potential. Again, no consistent change in amplitude or latency of either evoked potentials was noted.

Lai et al. [1988] measured the cholinergic activity in the brain tissue by determining the sodium-dependent high-affinity choline uptake (HACU) of brain tissue of rats after 45 minutes exposure to 2.45 GHz at 0.6 Wlkg whole-body average SAR in either pulsed radiation (500 pps, 2 J.ls pulse duration, 10-3 duty factor) or CW radiation in a circularly polarized waveguide exposure system or in the far-field in a miniature anechoic chamber. They found a decreased HACU in frontal cortex regardless of pulsed or CW radiations or exposure system used. Decreased hippocampal HACU occurred only after exposure to pulsed, but not CW radiation in both exposure systems. Decreased striatal HACU occurred in rats after exposure to either pulsed or CW radiation in the miniature anechoic chamber, but not after exposure in the circularly polarized waveguide. None off our exposure conditions altered the hypothalamic HACU. No significant pattern was found between brain local SAR and decreased HACU. This led the authors to conclude that the effect of microwaves on the brain HACU originated from other sites in the brain or body. Decreased frontal and hippocampal choline uptake, an indication of decreased cholinergic activity, was considered to be the mechanism for decreased working memory in pulsed exposed rats who committed more errors than sham exposed animals when tested in a radial arm maze [Lai et al. 1987, 1994]. The experiment on the pulsed effect has not be been repeated independently by other investigators.

Effects on Blood, Hemopoietic (Hematopoietic) System and Serum Chemistry

Czerski et al. [1974] and Baranski and Czerski [1976] reported a study aimed to compare the effects of2.95 GHz CW (n= 5) and pulsed (n= 9) microwaves on the hemopoiesis of rabbits at 30 W/m2 averaged power density, 2 hours daily for 74 hours. Characteristics of these microwave pulses were 1 J.lS pulse duration, 1,200 pps and 1.2 x 10-3 duty factor. In comparison

to the control group (n=19), erythrocyte concentration, hemoglobin concentration and hematocrit did not alter significantly in rabbits exposed to CW or pulsed microwaves [Baranski and Czerski 1976]. On the other hand, alterations in iron kinetics were noted. These changes included increased half life, decreased iron transport rate, decreased iron turnover rate, decreased iron incorporation into erythrocytes and decreased percent of erythrocyte production [Czerski et al. 1974]. Czerski et al. [1974] concluded that perhaps the most interesting finding is that 74 h of

228

exposure to pulsed microwaves induced much more pronounced effects than to CW of the same dumtion, the difference between these groups being highly significant. The overall picture of the alteration in iron kinetics appears to be consistent with a slower turnover of circulating

erythrocyte, perhaps erythrocytes lived longer in exposed than in control rabbits. However, overlapping of error bars was noted. Due to the absence of details in inferential statistics and the nature of the error bars, it is difficult to reconcile the validity of a pulsed effect.

Chou et al. [1982b] in the previously referred study also compared the effects CW and pulsed microwave had on blood cellular elements. Hematological endpoints were red blood cell concentration, white blood cell concentration, hemoglobin concentration, hematocrit, mean corpuscle volume, mean corpuscle hemoglobin concentration, % of reticulocyte and differential leukocyte (neutrophil, lymphocyte, eosinophil, basophile and monocyte) concentrations. These endpoints were evaluated either at monthly intervals or at the end of the three month exposure. Tabulated data (mean and standard deviation) were presented along with the "normal range" reported in the literature for this species of animal. It was concluded that minor morphological

alterations of red blood cells (poikilocytosis - irregular shape, polychromasia - stained with basic and acid dyes, anisocytosis - considerable variation in size) occurred in all three groups of animals. No statistical differences were found among them. Exactly the same conclusion was presented by Baranski and Czerski [1974]. Because of large biological variations and subtle differences

between the effects of CW and pulsed microwaves, the authors [Chou et al. 1982b] concluded that a larger number of animals should be used in future experiments.

Baranski [1971] compared the effects ofCW and pulsed 3 GHz (10 cm) microwaves at 35 W/m2, 3 hours daily exposures, except Sundays, for 3 months on the hemopoietic system. The pulse characteristics were not specified. A relatively large number of animals (n= 50), including rabbits and guinea pigs, were used in each sham, CW and pulsed exposure. At the end of 3 month exposures, red blood cell concentration did not change significantly from those of sham exposed animals in both species of animals exposed to CW and pulsed microwaves. On the other hand, the author reported alterations of erythroblast maturation process in bone marrow of guinea pigs exposed to microwaves. Immediately after 3 months of pulsed exposure, a marked depression in the percentage of proerythroblasts and basophilic erythroblasts was noted. This depression was accompanied by increases in percentage of polychromatic and orthochromatic

erythroblasts indicating an accelerated maturation process. Similar changes in the erythroblast maturation process was noted but to a lesser degree in guinea pigs exposed to CW than to those exposed to pulsed microwaves. However, changes in erythroblast maturation did not result in a change of the total erythroblast percentage in bone marrow cells. In addition, the mitotic index of the erythroblasts in bone marrow decreased after exposure for 3 months, even more after 4 months. One month after termination of exposure to CW the mitotic index returned to normal, but overcompensation was noted one month after pulsed microwave exposure. The mitotic index reached nearly twice the control value in pulsed exposed guinea pigs. Since circulating red blood cell concentration was not evaluated 1 month after exposure, the possibility of resultant erythropenia and erythrocytosis is not known. Results of cytological examination indicated profound changes in the nuclei of erythroblasts. These changes included karyolysis, fragmentation of nuclei, pycnosis, clumping of chromosomes, chromosomal bridges and splitting off of chromosomes. However, quantitative or semiquantitative data was not presented.

Baranski [1971, 1972b] also noted increased white blood cell concentrations (leukocytosis) by approximately 50% to more than 100% in both species exposed to either CW or pulsed microwaves. This leukocytosis was contributed primarily by lymphocytosis (increased

229

lymphocytes). Lymphocytes increased further during the first 2 weeks and showed a tendency to nonnalization from the third week after termination of the 3 month exposures in guinea pigs. Difference between CW and pulsed microwaves on leukocyte concentration was not significant. In fact, leukocyte concentration in rabbits exposed to pulsed microwaves was consistently lower than those exposed to CW microwaves in both species of animals. The leukocytosis (lymphocytosis) was not accompanied by significant changes in percent composition of bone marrow cells. The stamped smear preparation of lymph nodes and spleen showed a marked increase in lymphoblasts and reticulum cells in exposed guinea pigs in comparison with controls. Abnormality of nuclear structure and mitosis were analogous to those seen in erythroblasts [Baranski 1971]. Mitotic rates oflymphopoiesis were further studied by the percentage of cells labeled with tritiated thymidine. Several fold increases in S-phase labeling were noted in lymph node and spleen of microwave exposed rats. Again, CW appeared to be more potent than pulsed microwaves in inducing increases in mitotic rate of lymphoblasts. The overall response of lymphocytosis appears to be an immuno-stimulation of unknown nature. In these experiments, multiple post exposure data was compared to a single control datum [Baranski 1971, 1972b]. Apparently, the control data were either pooled or historical in nature. If concurrent controls were not used, one has to question the validity of these results.

Wangemann and Cleary [1976] compared the effects of2 hour exposures to 2.45 GHz CW and pulsed microwaves on blood chemistry of Dutch rabbits. Two power densities, 100 and 250 W/m2 average were used. For pulsed microwaves, a 10 j.lS pulse duration and a fixed peak power density of 4.85 kW/m2 were used. Thus, two duty factors (0.0206 and 0.0485) and two pulse repetition rates (2.06 x 103 and 5.15 x 1 (j pps) were used. Additional groups of rabbits were exposed to 50 W/m2 and sham-exposed controls (n= 12). Five to six rabbits were used in all other groups. Blood chemistry endpoints were serum concentration of calcium, inorganic phosphate, glucose, blood urea nitrogen (BUN), uric acid, cholesterol, total protein, alkaline phosphatase (AP), lactic dehydrogenase (LDH) and serum glutamic oxalic transaminase (SOOT). These endpoints were obtained as pre-exposure baselines and post-exposure levels immediately, 1, 3, and 7 days after exposure. A significant rise in mean serum glucose concentrations was found in all microwave-irradiated animals immediately after exposure; and returned to baseline levels 1 day later. The magnitude of changes in serum glucose concentration was dose-dependent in CW exposed rabbits, 18% in 50 W/m2 CW, 291'10 in 100 W/m2 CW and 44% in 250 W/m 2 CW exposed rabbits. Similar changes at a lesser degree were noted in pulsed exposed rabbits. The increase in serum glucose concentration was considered to be a consequence of "thermal stress" because significant increases in rectal temperature were noted in rabbits exposed to 250 W/m2

CW and pulsed microwaves. In addition, significant increases in BUN (50%) and uric acid (150%) concentrations were noted in 250 W/m2 CW but not pulsed exposed rabbits. Focal hemorrhagic lesion and mild to moderate tubular nephrosis were noted in representative rabbits 24 hours after "2-3" hour exposure to 250 W/m2 CW microwave. Thus, gross and histopathological findings confirmed the observation of renal injury indicated by BUN and uric acid. Other focal lesions were observed in the psoas muscle, liver, kidneys, pulmonary tissue. However, no pathology was observed in animals examined 1-2 weeks after exposure. All other serum chemistry endpoints did not change significantly. The renal injury in the 250 W/m2 CW exposed rabbits appeared to be caused by a significant degree of hyperthermia since the rectal temperature of these animals increased by an average of2.7 °C reaching 40.7 °C at the end of a 2 hour exposure. The absence of renal effect in the 250 W/m2 pulsed exposed rabbits could also be explained by a lower degree of hyperthermia (1.7 °C increase and 40.1 0c) at the end ofa 2

230

hour exposure. Because of a large difference in degree of hyperthermia between CW and pulsed exposed rabbits, it is doubtful that the investigators administered an equivalent SAR during the CW and pulsed exposures.

Chou et al. [1982b] also compared the effects ofCW and pulsed microwaves on blood chemistry but at a lower intensity. As stated previously, the microwave was 2.45 GHz at 15 W/m2 average, 2 hours per day for 3 months. The blood chemistry endpoints were sodium ion concentration, potassium ion concentration, chloride ion concentration, carbon dioxide concentration, ion gap, total protein concentration, blood urea nitrogen concentration, glucose

concentration, total complement concentration, triiodothyronine resin uptake, thyroxine concentration, percentage of albumin, and cortisol concentration. No change in any of these blood chemistry endpoints was noted. These results seem to support the thermallhyperthermic nature of the changes in blood chemistry of the Wangemann and Cleary [1976] study.

Lymphoblastoid Transformation

A study by Stodolnik-Baranska [1967] has raised the possibility that the

immunocompetent cells of humans are particularly susceptible to microwave radiation. In fact, it was claimed that microwaves were the only physical agent known to cause stimulation of spontaneous lymphoblastoid transformation. These studies were admitted by some authors to be poorly reproducible and nonquantitative [Roberts et al. 1983]. Nonetheless, the study is frequently cited, and has provided the limited data available on exposure of human leukocytes, for use by individuals and agencies that develop environmental health standards [Roberts et al. 1983].

Stodolnik-Baranska [1974] summarized the stimulatory effect of pulsed 2.95 GHz microwaves on spontaneous lymphoblastoid transformation of human lymphocytes in culture. The mitotic index of human lymphocytes in culture was used for identifying the presence of spontaneous Lymphoblastoid transformation. Pulse characteristics were 1 ji-S pulse duration,

1,200 pps and 0.83 x 10-3 duty factor at 70 and 200 W/m2 average power densities. Mitotic index

of cultured human lymphocytes was found to increase to approximately 2 times that of controls after 20 and 40 minutes of exposure but not after 5, 10 and 15 minutes of exposure at 200 W 1m2•

Similar results were obtained in human lymphocyte cultures after 3 or 4 hours exposure at 70 W/m2, but data were not presented. When cultured human lymphocytes were exposed at 200 W/m2 for 10 minutes, the increased mitotic index became noticeable after 59 hours, became pronounced (approximately 40 % increase) after 64 hours, and subsided to a normal rate after 70 hours in culture. A 0.5 °C increase in culture temperature was noted after a 15 minute exposure

at 200 W/m\ and 1.0 °C after 20 minutes at the same intensity. The medium temperature remained constant during a 4 hour exposure at 70 W/m2. Increased medium temperature is known to cause enhancement of the spontaneous lymphoblastoid transformation [Czerska et al. 1992] as shown in Fig. 4. Other changes noted by Stodolnik-Baranska [1974] were increased

chromosome abnormalities (2 to 17 times that of controls) in human lymphocyte culture exposed at 200 W/m2 for 5 to 20 minutes. These chromosome abnormalities included stickiness and chromosome aberrations (dicentrics, hypoploidy, hyperploidy and breaks).

Using a waveguide exposure system [Lu et al. 1983], Roberts et al. [1984] studied human

lymphocytes exposed to 2.45 GHz microwave pulses modulated at 16 and 60 pps at 0.39 and 4 W/kg for two hours. A relative high duty factor, 0.5 was used. Unstimulated and phytohemagglutinin suboptimally and optimally stimulated tritiated thymidine and tritiated leucine

231

incorporations for DNA and total protein syntheses were used to identify spontaneous lymphoblastoid transformation, and activated lymphoblastoid transformation. Viability of lymphocyte was also evaluated. Lymphocytes from individual donors were split into two parts for sham- and microwave exposures to limit the influence of individual variation on the outcome of the experiment. Two identical waveguides, one energized and one not, were mounted on a platform connected to a reciprocating shaker to provide agitation and to prevent sedimentation during exposures. Therefore, the temperature distribution within the exposure vials was much more uniform than in those exposure systems without agitation. The same exposure system, procedure and endpoints were also used to study the CW effect on lymphoblastoid transformation [Roberts et al. 1983] at 0.5, 1.0 and 4.0 W/kg for 2 hours with the exception of an additional endpoint, tritiated uridine incorporation for RNA synthesis. Roberts et al. [1987] also compared the effectiveness ofCW and pulsed (0.5 duty factor) 2.45 GHz microwaves at 4 W/kg on the unstimulated and mitogen-stimulated responsiveness of lymphoblastoid transformation in influenza virus infected human lymphocytes. None of these studies showed any evidence of a difference in any of the endpoints either unstimulated or stimulated between control and exposed split samples. The lack of a demonstrable difference was not originated from a lack of responsiveness of human lymphocytes used because these lymphocytes responded to mitogen stimulation in a dose dependent fashion regardless of their exposure history. Using the same procedure, Roberts et al. [1985] demonstrated a sensitivity to microwave induced effects in lymphocytes exposed to 2.45 GHz CW at 22.5 W/kg for 2 hours (culture temperature= 42.7 0c). This exposure resulted in decreased unstimulated RNA and total protein syntheses, as well as delayed synthesis of DNA, RNA, and total protein in response to stimulation with the optimal concentration of mitogen and decreased synthesis in response to suboptimal concentrations of mitogen.

On the other hand, Czerska et al. [1992] provided convincing evidence ofa profound stimulatory effect of pulsed 2.45 GHz on the sponntaneous lymphoblastoid transformation of cultured human lymphocytes (Fig. 4). A shorted rectangular waveguide exposure system in an incubator at 37°C was used for this experiment. The CW exposures were a "non-heating" level

«0.2 °C) at 0.8-1.3 W/kg, and several "heating" levels (0.5,1.0, 1.5 and 2.0 °C increase) at 1.8-

2.3,3.5-4.5,6.8-8.3 and 9.8-12.3 W/kg. The pulsed microwaves were operated at 1 f.,lS pulsed duration, 100 to 1,000 pps, and variable duty factors, 10-4 to 10-3. It appears the peak SAR were fixed at 9.8-12.3 kW/kg. The exposure duration was 5 days continuously. Because the lymphoblastoid tranformation process involves gradual enlargement of the cell and of the nucleus, the size (>200 f.,lm2 vs <100 f.,lm2) of the cell was used to identify lymphoblasts, lymphocytes and

intermediate cells at the end of 5 day experiment period. In addition, conventional heating controls were used at 0.5 °C intervals between 37 and 39°C for 5 days. Figure 4 shows clearly the optimal temperature for lymphoblastoid transformation was 38.0 °C irrespective of the exposure. The number of blastoids formed was not different between lymphocytes exposed to conventional (ambient) heating or CW microwave. However, the lymphoblasts formed from PW exposure were several times greater. Due to the known sensitivity oflymphocytes to increased temperature, inhomogeneous distribution of culture temperatue may have caused the cell destruction in microwave exposed human lymphocytes at temperatures as low as 38.5-39.0 °C. The temperature inhomogeneity could not fully explain the difference in lymphoblastoid transformation rates between CW and pulsed microwaves. Additional experiments are needed to confirm this pulse enhancement effect.

232

25.---~,-----~------.-------.------.----~

20

g 15

~ U)

:'5 m 0 10 I n. :::!: ~

5

0

o AMBIENT ~ CWo 1.2.4.7.5 &< 11 W/kg -3

~ PW. 1 #S. 100-1.000 PPS. 10 D.F.

RANGE= S.D.

z z 0 0 ~ ~ u u :::> :::> rr rr l- I-U) U) W W 0 0

...J ...J

...J ...J W W U U W

...J Z

...JO « ~o J i= Irl-

O~~ «0 -,,-0. Z

37.0 37.5 38.0 38.5 39.0

INCUBATION TEMPERATURE

Figure 4. Effect ofCW and puJsed microwaves on spontaneous lymphoblastoid transfonnation of human lymphocytes in vitro. The graph is ploted from data presented by Czerska et al. [1992]. The lymphoblasts were cells whose size is larger than 200 J.lm2 at the end of 5 day experiment. The incubator temperature was maintained at 37°C during 5 days of CW and pulsed microwave exposures. The "0" at 38.5 ·C indicates that no lymphoblast was found.

Other Studies

Sanders et al. [1985] compared the efficacy of 5 minute exposures of two 591 MHz microwaves (250 and 500 pps, 5 fJ.S pulse durations) on the brain concentration of nicotinamide

adenine dinucleotide reduced form (NADH) in rats. They found a factor of 4 increase in thresholds, from about 4.5 to 18.5 W/m2, when pulse modulation frequency decreased from 500 to 250 pps. The effect of CW exposure on brain NADH concentration was not included. Thus, the specificity or moditying effect of pulse modulation on the studied endpoint could not be evaluated. In addition, they compared the efficacy among CW, 16-Hz sinusoidal-amplitudemodulated and 500 pps pulse-modulated (5 fJ.S pulse duration) 591 MHz microwaves on brain

adenosine triphosphate (ATP) and creatine phosphate (CP) concentrations at 138 W/m2 rms average (2.48 W/kg) for 0.5, 1 and 5 minutes. Both ATP and CP concentrations were significantly reduced from those of sham-exposed rats, but the efficacy of modulation was not evaluated due to large variations of the ATP and CP concentrations. Based on the absence of a brain temperature increase, the authors suggested a direct interaction of microwave energy with the mitochondrial electron transport chain function of ATP production.

Rafferty and Knutson [1987] compared the phase transition temperature of a liposome (dipalmitylphophstidylcholine and diapalmitylphosphtidylglycerol in 4: 1 ratio) while exposed to 0.93 GHz CW (300 W/kg average SAR) and pulsed (4,000 kW/kg peak SAR, 300 W/kg average SAR, 10 fJ.S pulse duration, 7.5 pps, 75 x 10-6 duty factor). The phase transition temperature of sham exposed liposomes was 40.1 ± 0.12 °C (S.D.). Neither CW nor pulsed microwave exposure altered this phase transition temperature.

233

Roberti et al. [1975] compared the effect of3 GHz CW and pulsed microwaves on the spontaneous activity of rats. The pulse characteristics were 1.3 fJS pulse duration, 769.23 pps and

a duty factor of 10-3. Sham-exposures for 185 hours, CWat 5-10 W/m2 for 185 hours, pulses

at 15-20 W/m2 average (15-20 kW/m2 peak) for 185 hours, and pulses at 240-260 W/m2 average (240-260 kW/m2 peak) for 408 hours were used. Spontaneous activity was not different between the sham exposed group and any of the microwave groups. This report indicates that spontaneous activity is not a sensitive measure for comparing effects between CW and pulsed microwaves.

McLees et al. [1972] compared the effect of 13.12 MHz CW and pulsed microwaves on regenerating hepatic tissue of the rat. The pulses were 200 fJS pulse duration, 500 pps, and 0.01

duty factor. A thermometric method was used to determine the whole-body averaged SAR. The averaged SARs were 1.2 to 1.3 W/kg for both CW and pulsed (120 to 130 W/kg peak SAR) microwaves. Power density or electric field intensity was not specified. Rats were hepatomized under ether anesthesia with removal of 60-70% of the original hepatic mass. Exposure was

initiated immediately following hepatectomy for 28-44 hours. Concurrent sham-exposed rats were used in conjunction with each of two microwave exposure protocols. Endpoints of the study were weight loss, mitotic index of the regenerating liver, and chromosome aberrations of the regenerating liver. Statistical differences between endpoints from either of the microwave exposed groups and their respective concurrent shams were not found. Obviously, regenerating liver seems not to be susceptible to microwaves either.

TEMPO EXPERIMENTS

The Transformer Energized Megavolt Pulsed Output (TEMPO) system at the Walter Reed Army Institute of Research (WRAJR) is a 3 GHz transmitter capable of delivering pulses (Fig. 5) at 500-700 MW peak output power, 120 ns wide, 20 ns rise time, and 0.125 pps. The transmitted power varies, a characteristic inherent in virtual cathode oscillators, yet can be represented as a rectangular pulse of 200 MW for 80 ns. With a 0.125 Hz pulse repetition rate, the duty factor is 10-8 and the average output power is 2 W. The TEMPO antenna is a longitudinally slotted circular waveguide which radiates a fan-shaped (narrow horizontal, wide vertical) horizontally polarized beam. Subjects are typically exposed in a dual comer reflector assembly, 2.25 m from the antenna. The comer reflectors provide a 10 dB enhancement and achieve 10 W/m2 from a 1 W/m2 free-field power density per watt transmitted power. Thus, at maximum transmitted power (700 MW), peak power density is 7 GW/m2 and the peak electric field intensity is approximately at 1,600 kVlm. Wire screens of varying mesh densities can be placed over the antenna aperture to attenuate output of the system. These screens produce power outputs of -20, -30, -40, and -50 dB. Calorimetry results indicated that the whole-body average SAR of an average rat was 0.036 W/kg per watt transmitted power or 0.072 W/kg at full power. Extensive local SAR of interest has also been evaluated by a thermometric procedure [Raslear et al. 1993b]. In addition to microwaves, TEMPO produces the noise and soft x-rays. For 200 pulses, the ionizing radiation dose was less than 0.31 rem, and noise level was lower than 89 dB SPL (57 dB during sham exposure). The TEMPO system has been used for studying the behavioral toxicology of high peak power but low average power microwaves. A typical exposure consisted of 200 pulses in approximately 25 minutes. Several studies have been performed at WRAIR with this device.

234

; II i I

I

I !

tn. I'! i i

200

ns

Figure 5. Oscilloscope tracing of a typical TEMPO pulse. Spikes and ringing are noted.

Raslear et al. [1993b] trained rats under a bisection paradigm in which the rats were trained to discriminate between two durations of light (0.5 and 5 s, training stimuli) to obtain food reinforcement with a response on the appropriate lever designated as short and long. Five new durations (0.74, 1.1, 1.62, 2.4, and 3.54 s, test stimuli) equally spaced on a logarithmic scale between training stimuli were added for testing. All seven stimuli were presented in random order throughout the test session (320 trials per session). Responses to the test stimuli were never reinforced, but responses to the training stimuli continued to be reinforced. Psychometric functions, bisection point, response bias, discriminabiIity, general task performance and null response (failure to respond on a trial) were examined after exposure to 200 TEMPO pulses at 5 graded doses in a 50 dB range. Both ascending and descending series of exposures were used. In addition, a cage control and a sham exposure were used. Measures of time perception, response bias, total trials and discriminability (ability to distinguish between durations) were not affected by TEMPO exposure. Significant dose-dependent changes were noted in increased session time, increased number of null responses and increased cumulative null responses with dose. It was interpreted that TEMPO exposures at less than 0.072 W/kg whole-body average SAR were affecting cognitive function in the rats, particularly the decision-making process. However, post hoc analysis was not performed to identifY a threshold for each behavioral endpoint.

Using a Y-maze, Raslear et al. [1991a] tested 24 hour memory consolidation in rats exposed to 200 TEMPO pulses at full power (0.072 W/kg whole-body average SAR), at 30 dB attenuated power, and in rats that served as cage controls. The difference between the two exposed groups was not significant. However, both exposed groups reliably made more errors than the cage control group. They concluded that the memory deficit caused by TEMPO pulses was consistent with the results of the bisection study [Raslear et al. 1993b] in which an increase in null responses, reflected a lack of confidence in making a decision.

Akyel et al. [1993b] studied the treadmill performance in rats immediately after sham or full-power TEMPO exposure (0.072 W/kg whole-body average SAR, 200 pulses). Cage controls

235

were also used. The running time of TEMPO exposed rats was reduced (56.44 ± 10.69 s, mean

± S.E.) from those of cage control (150.13 ± 25.65 s) or sham-exposed rats (152.89 ± 35.18 s).

The mechanism of this decreased endurance was not investigated. Raslear et aZ. [199Ib] also studied the effect of TEMPO pulses on motivation for food and

the circadian rhythmicity of food acquisition. Rats were exposed to one of three conditions, full power TEMPO pulses (0.072 W/kg whole-body average SAR, 400 pulses, 50 minutes), pulses at -30 dB (400 pulses, 50 minutes) and sham-exposure for 50 minutes. Following exposure, each rat was placed in one of six similar home cages which was equipped with a response lever and food pellet dispenser. On the first day in the test apparatus a lever pressing response produced a single 45-mg food pellet. On the second day the response requirement remained at 1 response/food pellet (FR I, Fixed Ratio 1). On subsequent days the response requirement was increased to 15,45,90, 180 and 360 responses per pellet. The animals were free to respond at all times of day. A 12: 12 light:dark cycle (light on at 0600, off at 1800 hours) was used. The only food was what the rats earned in the experiment. The number of food pellets consumed as a function of the price of the food (response requirement) followed the typical pattern of decreasing consumption as price increased. There were no apparent differences among groups. Circadian function was assessed by the number of food pellets obtained during the day and the night. Typically, rats consumed 90-100 % of their food during the night. With the exception of 1 pellet per response (FR 1), TEMPO exposures did not affect the circadian rhythmicity. Instead of consuming 90-100 % of their food during the night, both TEMPO exposed groups consumed only about 50% of their food during the night under the FR 1 schedule.

The Porsolt swim test [Porsolt et al. 1978] in rats was used to identify the existence of despair and depression after an exposure to full power TEMPO pulses (0.072 W/kg whole-body average, SAR, 200 pulses) [Raslear et af. 1993a] There was no difference in immobility time between TEMPO exposed and sham-exposed rats. It appeared that exposure to TEMPO pulses did not cause despair and depression in rats irrespective of other behavioral alterations caused by the same exposure.

D'Andrea et aZ. [1993] used the TEMPO exposure system for an operant behavior study in rhesus monkeys. The corner reflector was not used, therefore, the peak power density was 456

kW/m2 which resulted in a peak whole-body SAR of2.21 MW/kg and 1.3 J/kg SA per pulse. Performance under a color discrimination task for food was used as the endpoint. The task required monkeys to pull one plastic lever on a variable interval schedule (VI-25 s) and then respond to color signals and pull a second lever to obtain food. Behavioral performance on either

component of the task was not altered significantly by the TEMPO pulses. D'Andrea et aZ. [1989] also exposed monkeys to high peak power 2.37 GHz pulses from

a different TEMPO system. The pulse characteristics were 70-113 kW/m2 peak power density and 93 ns pulse duration. The estimated temporal-peak spatial-average whole-body SAR was 583-938 kW/kg or 54-87 mJ/kg per pulse. The number of pulses delivered was between 28 and

51 in 20 minutes. Therefore, the whole-body average SAR was estimated to be less than 0.1 W /kg. The endpoint of the study was performance of the monkey under a vigilance task. The task involved two components, responding on a variable interval schedule (VI 20 s, 1-84 s range) on one lever and responding on a second lever to tones. Compared to sham irradiation, significant changes in performance were not observed. In a subsequent technical report, D'Andrea et al. [1990] also found no effect of different TEMPO pulses (365-827 kW/kg peak SAR, wholebody average SAR < 0.1 W /kg) on the behavior of monkeys under the same vigilance task.

Klauenberg et aZ. [1988] exposed rats to TEMPO vircator pulses at 1.26 GHz with a 85

236

ns pulse duration and a peak power density of 5 kW/m2. Startle, following single and multiple pulses (1 pps, for 10 s), general activity, and rotarod performance after mUltiple pulses were used to assess the effects of exposure to TEMPO pulses. Reliably more animals in the exposed group startled relative to the sham-exposed group when single pulses were used, but no differences between groups occurred during multiple pulse exposures. More exposed animals failed to complete the rotarod task than sham-exposed animals.

Cordts et al. [1988] exposed rats to a Cober peak power simulator operating at 2.066 GHz, 250 ms pulse duration, and 300 W/m2 peak power density and a Gypsy Vircator pulse operating at 1.64 GHz, 140 ns pulse duration and 80 kW/m2 peak power density. Three behavioral tasks, one-trial avoidance, drinking, and rotarod performance, were used to evaluate the effects of the Cober and Gypsy pulses. Only the drinking test (time spent drinking) was reliably reduced by either source, and similar effects were seen for both pulses despite the large difference in incident power densities. It was estimated that a 0.7 °C increase in body temperature was caused by exposure to Cober pulses while only 0.0084 °C was caused by Gypsy pulses. Hence a conventional concept of heating (differential increase of body temperature of either 0.1 or 1 °C depending on endpoints) fails to explain the reduced drinking effect.

It appears that some cognitive behaviors are susceptible to the high peak but low average power microwave pulses. On the other hand, well trained behaviors appeared to be resistant to alteration by this type of microwave pulse. Unfortunately, none of these studies has attempted to include experiments using a CW microwave of the same carrier frequency and equivalent low average SAR exposure thereby limiting an evaluation of a pulse modification factor.

BIOLOGICAL EFFECTS OF CARRIERLESS PULSES

The current IEEE standard [IEEE 1999] recommends a 100 kV/m peak electric field intensity limit for pulsed fields oflow frequencies and short pulses. This recommendation for a peak E-field limit of 100 kVlm is based on the necessity to cap the allowable field below levels at which air breakdown or spark discharges occur. The IEEE Standards Coordinating Committee 28 [IEEE 1999] also recognizes the sparseness of studies for very short high-intensity exposure. This section is intended to update current knowledge in evaluating the biological effects caused by exposure to carrierless pulses in which a center frequency is not apparent. This type ofRF is usually produced by impulses instead of modulation ofa narrow-band RF.

Absorption of Carrierless Pulses

Guy [1990] used a Numerical Electromagnetic Code Method of Moments to calculate the maximum induced currents, SAR and absorbed energy density in human and non-human primates exposed to EMP. His results are shown in Table 4. The induced current could vary with size, as high as 281 amperes per leg for a human male, to as low as 4.89 amperes per leg for a squirrel monkey exposed to a 100 kVlm, 10 ns rise tittle EMP. However, the variation of maximum current density and SAR for the exposed subjects was small, ranging from a low of 148 kAlm2 and 26.2 MW/kg for an adult human, to a high of 180 kAlnr and 38.0 MWlkg for a 1-year-old human child. Absorbed energy per pulse and peak E-field in tissue were also relatively independent of size. The tissue peak E-field was approximately twice the peak E-fie1d intensity of the ambient (100 kV/m) EMP pulse.

237

Table 4. Comparison of maximum induced current, current density, SAR, absorbed energy and E-Field in human and non-human primates exposed to 100 kVlm, 10 ns Rise time electromagnetic pulses

Height Current Current SAR Energy E-field Subject per leg density per pulse in tissue

{cm} {A} {kNm2} (MW/kg) {J/kg} (kVlm}

Man 178 281 148 26.2 0.186 177

10-year-old-child 138 185 162 31.2 0.182 193

5-year-old-child 112 129 171 34.6 0.177 202

l-year-old-child 74 59.0 180 38.0 0.164 211

Rhesus monkey 40 16.3 173 34.4 0.149 200

Sguirrel monk!2: 23 4.89 155 28.4 0.142 183 * Data adapted from Guy [1990].

Additionally, Lin [1975, 1976, Lin and Lam 1976, Lin et al., 1975] using frequency analytic techniques for various electromagnetic pulse interactions with different biological models explored the dependency of transmitted EMP on models, on pulse duration and on wave shape. Lin and Lam [1976] indicated that the coupling efficiency for a one jJ-S Gaussion pulse is three times as big as that for a ten JJ-S Gaussion pulse in a model composed of a layer of soft tissue over a semi-infinite medium of bony structure. The maximum transmitted pulse strength was calculated to be about 2.2 kV/m from one /is pulse at 100 kV/m. A similar dependency of transmitted EMF on pulse duration was noted between 1 jJ-S (221 VIm) and 50 jJ-S (188 VIm) 50 kVlm Gaussian EMP incident normally at the air-tissue interface of a semi-infinite model [Lin 1975]. Lin [1976] concluded that the transmitted field strength in homogeneous spherical models of human and animal heads (10, 5, 3.5 and 1.75 cm radius) from 100 kV/m Gaussian EMP varying from 1 to 10 JJ-S pulse duration was quite small, less than 20 VIm was coupled into a 3.5 or 10 cm radius spherical head. On the other hand, a triple exponential 50 kVlm EMP could produce 735 VIm in a 10 em radius spherical head and 472 VIm in a 3.5 cm radius spherical head.

Using the same technique, Guy [1989] concluded that a peak current level of 5 A per kVlm occured for exposures to EMPs with a 10 ns rise time. Hagmann [1993] measured the peak current induced in human volunteers. Measured peak current levels per leg were 1.7 A per kV/m at the "ALECS" EMP facility and 0.9 A per kVlm at the "EMPRESS f' EMP facility. If both arms and opposite leg of volunteers were raised, the peak current was enhanced (2.95 A per kV/m) at the "ALECS" facility but less so (Ll A per kVlm) at the "EMPRESS f' facility. Clearly, the magnitude of current induced through the human body by EMPs can be greatly modified by changing posture and limbs making ground contact.

Chen and Gandhi [1991], using the finite-difference time-domain (FDID) technique, calculated currents induced in an anatomically based model of a human with rubber soles for exposure to vertically polarized electromagnetic pulses at a peak electric field of 55.4 kV/m. They noted an excellent agreement of the calculated results by FDID technique with data measured for a human subject. The peak current induced in both ankles, both knees and both

238

thighs (220 to 284 A by a 55.4 kVlm peak EMP) were in agreement with Guy's results (Table 4). On the other hand, the whole body average SA of two representativeEMP pulses (0.43 mIlkg from a 53.5 kVlm peakEMP pulse, and 0.22 mIlkg from a 41.5 kV/m peak EMP pulse) differed from Guy's results (Table 4) by 3 orders of magnitude.

For extrapolation from results of animal experimentation to human health, it is important to know which absorption metric is responsible for the biological effect. The size dependence of peak induced current by EMP pulses was also evident in rats, a smaller laboratory animal. Mathur et al. [1990] measured the induced current in a 200 g rat (body length - 15 cm) exposed to an EMP pulse (67 kVlm peak, 8 ns rise time). The measured peak current was 1.5 A (22.4 rnA per kVlm) which is 2 orders of magnitude lower than the peak induced currents in humans (Table 4). The low peak induced currents by EMP pulses were confirmed by Gao et al. [1992] using FDID methods and a 2/3-muscle-equivalent model of a rat.

Potential Mechanisms for Tissue Damage from Carrierless Pulses

Albanese et al. [1994] proposed four potential tissue damage mechanisms for ultrashort electromagnetic signals. They were molecular conformation changes, alteration in chemical reaction rates, membrane effects (electroporation) and thermal damage. These proposed potential tissue damage mechanisms apparently required relatively high peak electric field and pulse duration. Some of these effects can be induced by pulses with a peak electric field intensity of several hundred kVlm.

The work of Neuman and Katchalsky [1972] was cited to support the existence of conformation changes (partial unwinding) in which artificial poly Npoly U triplets; destabilized in acid buffer were exposed to 2,000 kV/m, 10 j.LS electromagnetic pulses. The peak electric field intensity in this study was higher and the pulse duration was longer than those pulses typically produced by impulse radars or EMP simulators. In addition, Merritt et al. [1995] countered that several items including that DNA in situ is inherently stabilized by histone, repressor and activator proteins, and secondary chromosomal structures, and that topisomerase enzyme activity is required to release the stabilized structure and allow normal DNA synthesis to occur. Such dynamics may not be relevant as potential damage mechanisms.

Albanese et al. [1994] estimated the change of chemical equilibrium in the presence of an electromagnetic field as 1% per 1 kV/m of electric field intensity. Based on the rate constant of chemical reactions (l08/s1M, relaxation time 10 ns or longer), Merritt et al. [1995] concluded that ultrashort electromagnetic pulses with ps rise time and ns pulse duration could not alter the chemical reaction by individual pulses. In addition, Merritt et al. [1995] considered thilt chemical reactions are ll&SOciated with the average energy deposition instead of a transient electromagnetic field.

Another damage mechanism considered by Albanese et al. [1994] was the permeation of cell membranes by electromagnetic pulses of hundreds ofkVlm and pulse durations of j.LS to ms. The consideration was based on Tsong's review [Tsong 1991] of electroporation, a process which describes collectively the effect of electromagnetic pulses on cell membranes resulting in leakage of ions and metabolites, fusion of cell membranes, and opening of pores. Reversible openings of protein channels occur with J.LS pulses, while irreversible channel openings are seen as the results of ms pulses. The process of extrapolation from a dilute in vitro cell suspension to a complex in vivo system is questionable at the present time. Tsong [1991] also concluded that the breakdown

239

potential of bilayer lipid membrane is in the 150 to 500 mV range when the field duration is in microseconds to milliseconds. This value translates into a field strength of 30 to 100 MY 1m if the tissue thickness is 5 nm. Electroinsertion of foreign materials into a cell is known to require a lower critical field than electroporation [Mouneimne et al. 1993]. Furthermore, the critical field intensity routinely used in electroinsertion depends on cell size. Neumann et aI. [1993] found that in situ pulsed electric fields in the range of200 to 400 kV/m and 1 J,lS to 50 ms could be used in the membrane electroporation method for transferring genes, nucleic acids and proteins into a cell interior as well as electro-release of cellular components into the medium. Similar to electroinsertion, the threshold for membrane breakdown decreases with increasing cell size. The irreversible breakdown of cell membrane is about 200 kVlm for blood cells [Zimmermann 1982, Hansson-Mild et al. 1982, Paile et al. 1995], 100 kVlm for fibroblasts, and 5 kVlm for muscle cells [Lee et al. 1988]. With the exception of muscle cells, these threshold peak electric field intensities are higher than the current IEEE standards on peak electric field intensity [IEEE 1999].

The last tissue damage mechanism proposed by Albanese et al. [1994] was thermal damage by bulk heating. If a threshold bulk heating or temperature increase is allowed to continue for a sufficient time, tissue damage will indeed occur and it is not unique to ultrashort electromagnetic pulses. On a molecular basis, Albanese et al. [1994] suggested that if the rate of energy removal from a target molecule is slower than its absorption from the electromagnetic field, significant heating of the individual molecule could occur during exposure, with little gross temperature change in the overall medium, causing highly localized damage. The plausibility of the proposed mechanism is not known. Additionally, it was not clear to what extent molecule damage becomes critical to body homeostasis and is manifested in disease. Based on the results of highly sensitive single pulse endpoints (microwave hearing, microwave induced whole-body movements, and microwave induced thermal sensation), the presence of a critical duration is clearly evident. Although the critical duration appears to vary with the endpoint or effect studied, the effect threshold is dependent on the pulse energy (SA per pulse) when the pulse duration is shorter than the critical duration. When the exposure duration is longer than the critical duration, effect threshold appears to depend on the dose rate (SAR). Adair [1991] analyzed the electric and magnetic fields in the human body generated upon exposure to external pulsed electric fields. He concluded that for peak external field strengths as high as 100 kV/m, the effects of the consequent internal electric fields on sensitive cell elements, such as the membrane, organelles, and the macromolecules that carry genetic information, are shown to be small compared with the normal thermal agitation of the elements. He further concluded such pulses cannot be expected to produce any biological effects at the cellular level.

On the other hand, Gailey and Easterly [1993] reviewed the internal current densities and electric fields induced in the human body during exposure to EMP fields to predict the resulting membrane potentials. Using several different approaches, membrane potentials of about 100 m V were predicted. These values are comparable to the static membrane potentials maintained by cells. However, the EMP induced potential persisted for only about 10 ns, a time constant far shorter than the latency of most biological reactions in response to stimuli. Based on a theoretical model ofEMP coupling on cables, Agouridis and Easterly [1989] had concluded earlier that electromagnetic fields produced by EMPs and EMP simulators, in certain circumstances could be hazardous and/or exceed the present guidelines.

Gailey and Easterly [1993] also pointed out that the spherical cell model used in most studies is far from accurate. Muscle and nerve cells, for example, are elongated and the expression for limiting membrane potential must be modified by an enhancement factor. Other

240

considerations are electrical length of the structure, gap junctions electrically connecting groups of cells, altered capacitance of cell membrane, and thickness of membrane at tight junctions.

Using the Hodgkin and Huxley nonlinear membrane model, Bernardi and D'Inzeo [1984] studied the current density induced in the biological tissue on the resting potential of an excitable cellular membrane. Various characteristics of the incident field, such as wave-shape, pulse duration and amplitude were analyzed. For induced voltage lower than 1 mY, the membrane's behavior could be considered as linear. In this case, the membrane response showed a maximum response in the range of 1-30 MHz, where incident fields of about 20 kV/m are necessary to produce 1 mV on the membrane. Above 1 mY, the membrane showed a nonlinear behavior, and a CW incident field could produce a dc voltage on the membrane. Further increase in the amplitude of the incident field could produce depolarization or action potentials of the membrane. For pulsed fields, the membrane was particularly sensitive to pulses with a pulse duration of about 1 ms, the pulse duration comparable with the proper times of the membrane ionic channels. The threshold of action potential excitation varied inversely with pulse duration, i.e., the required incident electric field strength would be 400 kV/m at 1 ms and 2,000 kVlm at 10 f.,LS pulse

duration. They further commented that transmembrane voltages greater than 1 m V were produced by pulsed fields with 100 W/m2 average power density. But such pulses could not produce adequate transmembrane voltage to cause action potential excitation. On the other hand, specialized neuroreceptors and pacemakers might be more susceptible to such a transmembrane voltage. The validity of this potential mechanism has not been resolved.

Blair [1932] formalized the law of nerve excitation by deriving a mathematical equation based on the known resistive and capacitive properties of cell membranes. The Blair expression for the strength-duration curve is:

Where I is the peak current (or peak external or tissue E-field intensity) for a rectangular pulse of duration d; b is the rheobase and "t is membrane strength-duration time constant for the particular tissue. The "rheobase" can be described as minimum excitation threshold for longduration stimuli. For pulses shorter than the strength-duration time constant, the peak current/field intensity increases according to the duty factor, i.e., ratio of pulse duration to strength-duration time constant, for nerve excitation. The time constant could be identified as the break point between the minimum plateau and rising phases of the nerve excitation threshold curve in relation to pulse duration. The time constants for various excitable membranes (22 to 214 ~s) have been complied by Gedders and Baker [1989]. Reilly [1998] proposed a strengthduration time constant and "rheobase" for nerve excitation at 120 f.,LS and 6.2 Vim (in situ E-field).

The Blair relationship appears to hold true for pulse durations down in the 1 ~s range. If the Blair expression holds true for EMP and UWB pulses, the nerve excitation threshold intensity is 75 kV/m tissue peak E-field intensity for a 10 ns EMP pulse, and 750 kVlm for 1 ns UWB pulse. Apparently, the nerve excitation threshold by EMP pulses estimated from the Blair expression is about 37 to 42 % ofthe calculated tissue E field from a 100 kV/m EMP pulse (Table 4). In addition, the Blair expression predicts that the shorter the membrane time constant, the lower the nerve excitation threshold. Thus, because of the shortest time constant, a myelinated A fiber would be the most sensitive neural fiber to the influenceofultrashort pulses. Jokela [1997], using a SENN (Spatially Extended Non-linear Node) model with folded axon with 2 mm separation for

241

Ranvier nodes, calculated the nerve stimulation threshold for rectangular pulses. The asymptotic threshold level for electric field intensity was 10 Vim for pulse durations longer than 100 JiS, and a time-integrated electric field intensity of 1 x 10-3 Vim s for pulses with a pulse duration less than 100 JiS.

For nerve stimulation by sinusoidal currents, thresholds of excitation rise approximately in proportion to frequency beyond a critical frequency, a few kHz for neurosensory or neuromuscular stimulation, provided that the stimulus consists of multiple sinusoidal cycles [Reilly 1991]. Above this critical frequency, threshold current density for nerve excitation becomes so great that it will exceed the threshold for perception due to heating. For sinusoidal currents that are at least a significant fraction of 1 s in duration, the crossover between electrical and thermal perception thresholds occurs at about 100 kHz [Dalziel and Mansfield 1950, Chatterjee et al. 1986]. An experiment with anesthetized rats which allowed significant tissue heating, demonstrated neuromuscular stimulation with approximate frequency-proportional thresholds to 1 MHz [Lacourse et al. 1985]. Above this frequency, due to membrane short-circuiting, the heating of tissues by induced electric fields becomes increasingly important and tissue heating becomes predominant at frequencies above 100 MHz. In the frequency domain, pulsed fields are equivalent to a train of sinusoidal fields consisting of harmonic frequencies. Due to a very short pulse duration and even shorter rise times, the frequency spectrums of carrierless EMP and UWB span both non-thermal and thermal domains in their interactions with biological tissues. Therefore, entirely different modes of reaction may be encountered between pulses of similar peak electric field intensity and pulse energy.

Studies on the Biological Effects of Electromagnetic Pulses

There has been widespread public concern over the perceived risk ofEMP. During the latter part of the 1980s, lawsuits by several employees ofEMP test programs who had leukemia have caused the prohibition of many outdoor EMP simulator operations and increased public and political awareness of the EMP safety issue. The safety issue ofEMP needs to be addressed in a scientific manner since public concern about the potential biological effects ofEMP has been based on anecdotal evidence from studies without quantitative dosimetric data. Studies of biological effects of electromagnetic simulator pulses were performed as early as the 1970s, although defense contractors have been testing methods to protect electronic devices against effects of EMPs since the mid-1960s. However, only a limited number of studies has been performed.

The operation ofEMP simulators is primarily in the area of testing the vulnerability of electronic equipment to electromagnetic pulses generated by nuclear detonation. Concern for cardiac pacemakers in EMP fields falls under the same category of electronic vulnerability to EMPs. Patrick [1993] reviewed scientific literature in this area and concluded that there was little evidence of permanent damage from a single electromagnetic pulse up to 50 kVlm peak electric field intensity on cardiac pacemakers. A single failure was reported at a peak E-field intensity at 500 kVlm, however the failure could not be repeated after the damaged component was replaced. Cardiac pacemaker upset, induction of an additional beat or skipping of a beat, was observed in one unit at 1.25 kVlm for 2 % of the time. Two recent analytical and experimental studies were conducted at Harry Diamond Laboratories [Bock 1988, Ellis 1991] on three to ten modem cardiac pacemakers to "VEMPS IT' and "AESOP" electromagnetic pulses. Results of these studies indicated that a modern cardiac pacemaker would not experience damage or upset when

242

exposed to EMP simulator pulses up to a 16.5 kV/m peak electric field intensity, the typical field intensity in the accessible areas of these simulators. Apparently, the upper limit of cardiac pacemaker vulnerability to EMP pulses cannot be adequately tested prior to implantation in

patients. Therefore, Patrick [1993] recommended, as a caution, restriction of cardiac pacemaker wearers from EMP simulator areas and posted warnings of the presence of electromagnetic energy.

In 1985, the diagnosis of chronic myelogenic leukemia in one employee of an EMP testing program during a mandatory health surveillance examination prompted Muhm [1992] to perform

an epidemiologic study. Candidates for the study were identified from records indicating eligibility for undergoing mandatory health surveillance examinations because of employment in the EMP program between the inception of the surveillance examinations in September 1970 and an arbitrarily selected cutoff date, December 31, 1986. The surveillance examination was

typically required for anyone who was expected to be exposed to EMP for at least 30 days, not necessarily continuously, during a 6-month period. A total of352 candidates from 15 separate company documents was identified, and 304 men were followed for a total of3362 person-years. Standardized mortality ratios (SMR) were ascertained by two methods: the World Health Organization'S underlying cause of death algorithm; and the National Center for Health Statistics' algorithm to identifY multiple listed causes of death. There was one underlying cause of death due to leukemia compared with 0.2 expected (SMR= 437,95% confidence interval, CI= 11-2433), and two multiple listed causes of death due to leukemia compared with 0.3 expected (SMR= 775, CI= 94-2801). It was concluded that the study seem to suggest an association between death due to leukemia and employment in the EMP test program, but this suggestion was at most, tentative because of the absence of exposure data, the small number of deaths due to leukemia, and lack of statistical power and significance.

Sandler et al. [1975] studied the effect ofEMP on the morphology of freshly excised frog spinal cord and medulla in vitro. The EMP exposure duration was 25 ms (6 pulses) of 130 kV/m (44.8 MW/m~ peak electric field intensity pulses operated at 250 pps. The half width of the pulse was 30 ns. No difference in histology oflarge motor neurons between exposed and control tissue was noted. It was concluded that this EMP did not cause a non-thermal injury at a calculated temperature rise (0.001 0c) from each pulse. However, the tissue fixation occurred within 5

minutes including dissection of tissue and exposure. The time-frame of the experiment was too short for an appearance of morphological alterations in cells after such an insult.

Mattson and Oliva [1976] exposed a monkey for one hour (a total of 18,700 pulses) to EMPs at 266 kVlm (188 MW/m~ peak E-field intensity, 11 ns (10-90 %) rise time, 550 ns (lIe) fall time and 5 pps. Sidman avoidance behavior and post-exposure electroencephalogram were

evaluated. No difference in the endpoints between control and exposed sessions was noted. A series of chronic EMP experiments using rats, mice and dogs was performed at the

Armed Forces Radiobiology Research Institute. Characteristics of the EMP were 447 kV/m peak electric field intensity (530 MW/m~, 5 ns rise time and 550 ns (lie) fall time, and 5 pps. Initially, the exposure was 22 hr per day and 5 days per week for 38 weeks to a total of 108 pulses (note: 7.52 x 107 pulses by calculation) [Skidmore and Baum 1974]. Various endpoints were compared

between exposed and non-exposed animals. It was observed that the reticulocyte count in EMP exposed rats was nearly always greater than those in the non-exposed rats. However, there were no concomitant differences in peripheral erythrocyte count and in the radioactive iron incorporation between groups. Platelet counts in the exposed rats were about 10 % below those in the non-exposed group most of the time. Increased and decreased lymphocyte and total

243

leukocyte counts in EMP exposed rats were also found but less frequently than changes in platelet or erythrocyte count. When the EMP exposure was continued to 94 weeks to a total of2.5 x 108

pulses (note: 1.86 x 108 pulses by calculation), hematological changes were no longer evident

[Baum et al. 1976]. In addition, bone marrow cellularity was not different between groups [Skidmore and Baum 1974]. Because of these inconsistencies among hematological and hemopoietic endpoints and the absence of effect at the end of the 94 week exposure, it was concluded that EMP effect was not hazardous to rodents [Skidmore and Baum 1974, Baum et af. 1976]. In a later publication, Baum [1979] did not find any effects on hematological and hemapoiesis endpoints in four male and five female beagle dogs exposed to the 447 kVlm EMP pulses for 8 hr/day, for a total 45 days and 5.8 x 106 pulses (note: 6.48 x 1(1' pulses by

calculation). These endpoints were examined periodically during the 45 day exposure and 1 year after the conclusion of all exposures.

Other endpoints studied were chromosomal aberrations of bone marrow cells, blood chemistry, histology, teratogenesis, incidence of mammary tumors of rats, and leukemia incidence in leukemia prone AKR/J male mice. Differences betweeJ]. groups were not noted in any of these endpoints after a 38 week exposure [Skidmore and Baum 1974] and after a 94 week exposure [Baum et al. 1976]. In addition, fertility and reproductive capability were not affected by EMP exposure [Baumetal. 1976]. Teratogenesis and disturbance of reproductive capabilities ofEMP exposed male dogs were not found [Baum 1979].

Cleary et al. [1980] studied various biological endpoints in rabbits exposed to EMPs. The characteristics ofEMP were 100 to 200 kV/m peak electric field intensity, 0.1 f.)-S rise time, and an exponentially cosine decay pulse with 50 % pulse duration at 0.4 f.)-S. Because the spark

gap generator was free running, pulse repetition rates varied. They studied the sodium pentobarbital induced sleeping time in rabbits exposed to 140 kVlm at 10 pps, 190 kVlm at 24 pps, 90 kVlm at 10 pps. In the study of EMP effect on serum components (i.e., serum chemistry), 150 kV/m at 38 pps for 2 hours was used. Serum components were concentrations of alkaline phosphatase, lactic dehydrogenase, serum glutamic oxaloacetic transaminase, calcium ions, inorganic phosphate, glucose, blood urea nitrogen, uric acid, cholesterol, total protein, albumin, and total bilirubin, immediately and 24 hours after EMP exposure. For serum concentrations of triglyceride and creatine phosphokinase isoenzymes, rabbits were exposed to 200 kVlm at 50 pps and sampled immediately after the EMP exposure. In all of these biological endpoints, consistent statistical differences between EMP- and sham-exposed rabbits were not found.

Moore [1993] studied the orientation behavior of captured migratory birds in response to "EMPRESS IT' EMP pulses. The orientation of migrating birds was studied for 2.5 hours after exposure to a single 50 or 100 kV/m "EMPRESS" pulse. Rise time, fall time and pulse duration were not specified. The distribution of heading was found to be different between exposed and sham-exposed birds in one of three observation periods during the spring of '92 but not during the spring or fall of '91. The author noted that some individual birds may experience difficulty selecting a seasonally appropriate direction following exposure to a single EMP pulse at peak electric field intensity higher than 50 kV/m.

Moore [1993] also studied the natural remanent magnetization and isothermal remanent magnetization of heads of three bird species exposed to an ascending series of"EMPRESS If' pulses at 2, 5, 10, 19, 41, 72 and 102 kVlm. Magnetic field intensities and other pulse characteristics were again not specified for the "EMPRESS If' pulses. No effect ofEMP pulses was noted. Therefore, it was concluded that a discemable effect on magnetic orientation of bird

244

species due to EMP pulses was not expected to occur. Akyel et al. [1990, 1992] utilized a spectrum of behavioral paradigms (memory

consolidation, general motor activity, behavioral despair and preference of electromagnetic pulses) in rats in a study of the behavioral effects ofEMPs. The EMP exposures were performed in a parallel plate EMP simulator which produced electromagnetic pulses at 100 kV/m peak electric field intensity, 7 ns rise time, 20 ns pulse duration and 6 pps. The exposure duration was 20 or 30 minutes. Sham-exposed animals were used as controls.

In contrast to significantly more errors made by the TEMPO (1,600 kV/m, 20 ns, 3 GHz pulses) exposed rats [Raslear et al. 1991 a], the EMP exposed rats were not different from shamexposed rats when evaluated under a 24-hour memory consolidation test [Akyel et at. 1990].

However, visual observation and video recordings of the animals in the EMP simulator revealed that the EMP-exposed animals were more active than sham-exposed animals. In addition, exposed animals also displayed an ear "twitch" to most EMPs. During the Porsolt's swim test, the immobility time of exposed animals after a 20 minute EMP exposure were not different from those of sham-exposed animals [Akyel et al. 1992]. It was concluded that this EMP exposure did not affect the mood of animals immediately after exposure.

Preference of rats for exposed versus non-exposed areas was also studied by Akyel et al. [1992]. An 80 cm long Plexiglas tube was placed with a 40 cm section inside the EMP simulator and subjected to the EMP exposure and the other joining 40 cm section outside the EMP simulator out of the EMP field. For sham-exposure, the Plexiglas tube was placed in the same environment without the EMP field. A video camera was used to record the movement of subjects within the tube. Total elapsed time that the animal spent on either side of the tube was recorded for EMP- and sham-exposed animals [Akyel et al. 1992]. They observed that animals preferred the simulator side of tube when the simulator was active. The ear "twitch" was observed in exposed animals regardless of the position in and out of the EMP field and the "twitch" was not found in the absence of the field in sham-exposed animals. It was concluded that the "ear twitch" was not a field specific effect, rather, it was an effect caused by the noise associated with the generation ofEMP pulses

From these studies, it appears that EMP is not biologically active at peak electric field intensities below 100 kV/m. However, evidence of a transient hematological disturbance, startle, and slowing of memory process due to EMPs at 400-600 kV/m peak electric field intensity have also been observed. These experiments with positive findings need to be independently verified.

In addition, extrapolation from rodents to humans is hindered by uncertainties in mechanisms of action, and in determining an appropriate dosimetric index.

Studies on the Biological Effects of Ultra-Wi de-Band Pulses

The Ultra-Wide-Band (UWB) radiofrequency (RF) radiation is a new modality in radar technology and electronic warfare. Some of these sources can produce repetitive ns pulses with sub-nanosecond rise time and peak electric field intensity in several hundred kVlm [Agee et al. 1995]. At much lower power, an example offorthcoming civilian UWB radar application is in the automobile device for obstacle avoidance. Because the primary military purpose of high peak power UWB transmitters is to damage or destroy electronic equipment [Fuerer 1991], the laymen's concern of health effects is multiplied by the apparent higher field intensity that may be encountered in their operation. Due to a lack of dedicated exposure apparatus for biological experimentation, studies on biological effects ofUWB pulses are primarily limited to research

245

institutes affiliated with military organizations. Only a handful of research papers has been published due to the novelty of the technology.

Pakhomova et al. [1997] studied the mutagenic effects of UWB pulses in yeast, Saccharomyces cerevisiae, strain D7. This strain of yeast carries specific gene markers ala., trp5-12Itrp5-27, ade2-40/ade2-119, and ilvl-92/ilvl-92 which make it feasible to detect certain mutagenic and recombinagenic events from appearance of the colony, reverse mutation of the

isoleucine requirement, and mitotic gene conversion to tryptophane. The UWB exposure was performed in a "Sandia" device, where UWB pulses generated by a triggered spark gap generator

were transmitted into a giga-transverse electromagnetic (GTEM) cell designed for· small animal and in vitro exposures. Three 1 ml yeast suspensions at 107 cells per ml in a polystyrene tube (120 mm height x 17 mm outer diameter, 1 mm wall thickness) were exposed simultaneously.

Three different exposures for 30 minutes were used, i.e., sham-exposure, low repetition rate (16 pps, 104 kV/m peak electric field intensity, 165 ps rise time, and 1 ns pulse duration), and high repetition rate (600 pps, 102 kV/m peak electric field intensity, 165 ps rise time, and 1 ns pulse duration). Colony forming ability, normal and aberrant colonies, including mitotic crossovers, segregant, revertant and convertants, were scored. None of endpoints was found to be altered by UWB exposures. Therefore, it was concluded that UWB exposure used in this experiment did not change the colony forming ability of yeast cells nor did it cause mutations or gene recombination.

Pakhomova et al. [1998], in a follow-up study, used the same D7 strain yeast, but pretreated them with 254 nm ultraviolet light (UV) at 100 J/m 2 in 2.5 ml of 5 x 106 cells/ml in a 36

mm diameter Petri dish. The UV -pre-treated yeast was then exposed to UWB pulses in the GTEM cell for 30 minutes at doses and techniques as in the previous experiment. There were three exposure conditions: sham-UWB, 16 Hz, and 600 Hz exposures. The experiment was performed in parallel according to UWB treatments, i.e., sham-UWB exposure vs control, low repetition rate UWB exposure vs sham-UWB exposure, and high repetition rate UWB exposure vs sham UWB exposure. Sham UV exposure was not included in the experiment. The UV pretreatment was found to be effective in reducing cell survival to 60-70 % and the survival rate was not altered by UWB exposures. The frequencies of segregation, mutations to the isoleucine independence, and gene recombination events, such as mitotic cross-over and conversion to tryptophan independence were remarkably similar between samples in the parallel experiments, as well as among sham, 16 Hz and 600 Hz exposed samples. It was therefore concluded that these UWB exposure did not cause changes in UV -induced mutagenesis and gene recombination.

Erwin and Hurt [1993] presented probably the first summary of various studies on biological effects of UWB pulses performed under a collaborative program between Phillips Laboratory (Kirtland AFB, NM) and Armstrong Laboratory (Brooks AFB, TX). Endpoints of studies were cardiovascular response in rats, equilibrium task performance of monkeys,

performance scores of the Functional Observation Battery (29 physiological and behavioral measures) in rats, serum enzymes, plasma electrolyte concentration, hemoglobin concentration in rats, brain C-fos protein activation in rats, and metrazole seizure activity in rats. The exposure was with a "Hindenburg-2" (H-2) device, which was developed in mid-1992 [Agee et al. 1995] and produced bipolar pulses at 250 kV/m (166 MW/m2) peak electric field intensity, 318-377 ps rise time and 5-10 ns pulse duration. Typical exposure was 2 min at 60 pps except the operant behavioral study in which rats were tested immediately after an exposure to 12,000 pulses in a "sweep" pulse repetition, starting at 1 pps and increased continuously to 400 pps in 1 min. They concluded that none of the endpoints was affected by UWB pulses. Some of these studies were

246

subsequently published in 1995 [Sherry et aT. 1995, Walters et aT. 1995]. Sherry et al. [1995] exposed six adult male rhesus monkeys (Macaca mulatta) to UWB

pulses produced by the "H-2" device under similar parameters as used previously. The

performance of monkeys on the "Primate Equilibrium Platform" (PEP) was used as an endpoint.

The estimated whole-body average SAR was 0.5 W/kg. A 0.3 W/kg whole-body average SAR

was estimated by Hurt [1999]. Each monkey was exposed to these UWB pulses twice, separated

by 6 days between exposures. Performance on the PEP was tested before exposure, at 1 hr after

exposure and at 24-h intervals thereafter for the next 5 days. Sham-exposed monkeys were not

included. The only significant effects were individual differences among subjects and time.

Therefore, it was concluded that UWB exposure had no effect on PEP performance of monkeys.

In addition to an absence of sham exposure or sham-exposed animals, these monkeys had served

in experiments on the effects of low doses of carbamates and/or organophosphates on PEP

performance. The authors assured that such experiments had no detectable carryover effects. It

is understandable that the non-human primate is a precious research commodity and the training

for PEP performance is difficult and time consuming. Nevertheless, the lack of sham-exposed

controls and re-use of animals from other experiments raise questions as to the repeatability of

these observations.

Walters et al. [1995] reported the results of rodents exposed to UWB pulses produced

by the "H-2" device. The same pulse parameters were used at a 250 kVlm peak electric field

intensity (reported as 17 GW/m2, should be 0.17 GW/m2). The estimated whole-body average

SAR was 20 mW/kg [Hurt 1999]. Various endpoints were evaluated including performance in

a swimming test 3 h after exposure, scores in a Functional Observational Battery one day before

and 2 h after exposure, serum enzyme concentrations (aspartate transaminase, alanine

transaminase, creatine kinase and amylase) at 2 or 48 h after exposure, plasma electrolyte

concentration and osmolarity at 2 or 48 h after exposure, and brain C-fos protein 2 h after

exposure. No differences between any of the endpoints from sham-exposed and UWB exposed

rats were noted.

Miller et al. [1999] injected rats intraperitoneally with a convulsant, pentylenetetrazol

(PT) at 99 % effective dose (ED99' 89.17 mglkg) to re-evaluate the effect of UWB pulses on drug-induced convulsions. Four different treatment groups were used: UWB exposed-PT

injected, UWB exposed-saline injected, sham-exposed - PT injected, and sham-exposed - saline

injected. The UWB and sham-exposure were performed 20 s after the PT or saline injection.

UWB pulses were produced by a "Kentech PBG3" device into a parallel plate transmission line.

The UWB pulses were 40 kVlm (4.24 MW/m1 peak electric field intensity, 176 ps rise time, 1.92

ns pulse duration and 1,000 pps for 2 minutes. The whole-body SAR was estimated as 16

mW/kg. As expected, there was a significant drug effect (PT vs saline) on mortality, however,

exposure effects (UWB or sham) and interaction between drug and exposure effects were not

found. Since saline injected rats did not convulse, latency to first convulsion response, latency

to full tonic-clonic seizure and duration of seizure were evaluated in PT -treated rats only. No

differences were noted in any of these three latencies.

Seaman et al. [1998] tested the influence ofa "Sandia" UWB (100-105 kVlm, 165-168

ps rise time, 0.97 ns pulse duration) exposure on nociception and spontaneous activity and

interaction ofUWB exposure on morphine induced analgesia and hyperactivity of the mouse. A

complex experimental design with eight different experimental groups was used. These

investigators used cage control, sham exposure, low repetition rate UWB (60 pps, 105 kV/m peak, 165 ps rise time, 0.97 ns pulse duration, and 3.7 mW/kg estimated whole-body average

247

SAR), and high repetition rate UWB exposure (600 pps, 100 kV/m peak, 168 ps rise time, 0.97 os pulse duration and 37 mW/kg estimated whole-body average SAR) for 30 minutes. Saline- and morphine-injected male CF-l (7.5 mglkg) and male C57BU6 (10 mgIkg) mice served as subjects. Nociception was evaluated by latency of front and back paw licking and "first response" of the mouse after being placed on a 49.5-50.5 °C metallic surface. Other than the effect of morphine, none of the nociceptive endpoints was found to be influenced by the UWB exposures. Interaction between UWB and morphine was not significant in all of the three nociceptive endpoints. Lack of UWB effects on nociception and morphine-induced analgesia was supported by additional experiments with CF-l mice with 15, 30, and 45 minute exposures to the high repetition UWB pulses. Spontaneous activity was measured in C57BU6 by the number of beam crossings in a 38.5 em square arena. Other than a morphine induced hyperactivity, UWB effect and interaction between UWB and morphine were not observed. Due to limits of the exposure system, higher doses were not tested.

Inhibition of nitric oxide synthase (NOS) has a morphine-like analgesic effect and can cause alteration in locomotor activities. In an extension of their original study on morphine and UWB, Seaman et al. [1999] studied the effects ofa NOS inhibitor, ~-nitro-L-arginine methyl ester (L-NAME) on manifestation of UWB effects in CF-IlPlus mice. Nociception and spontaneous locomotor activity were measured as before. The sham and UWB exposures (600 pps, 102 kVlm peak, 0.90 ns pulse duration, 160 ps rise time, and 37 mW/kg estimated wholebody SAR) were administered to saline-injected and L-NAME-injected (50 mg/kg) rats. Effects of L-NAME on the "first response", hind paw licking latency and spontaneous activity were clearly demonstrated but no additional effects were found. While L-NAME was capable of ¥ucing increased spontaneous activity in sham exposed mice, its effect was blocked entirely by the UWB exposure. In supporting the evidence that UWB pulses could affect nitric oxide production, Seaman et al. [1999] referred to an enhanced nitric oxide production by 41 % in stimulated murine macrophages incubated with potassium nitrate (GLN treatment) after UWB exposure [Seaman et al. 1996]. The nitric oxide produced was 9.14 mM in sham-exposed and 12.89 mM in UWB-exposed GLN treated macrophages. The exposure of these murine macrophages, RAW 264.7, was performed in a T75 flask covered with 235 ml of phosphate buffered saline in the "Sandia" system to UWB pulses operated at 72-95 kV/m peak, 198-216 ps rise time, and 1.01-1.07 ns rise time for 30 minutes. A dose estimate was not included.

The study of cardiovascular effects ofUWB pulses has been performed in several studies. However, two different approaches have been used [Jauchem et al. 1998, 1999, Lu et al. 1999].

Jauchem's group used ketamine-anesthetized (150 mg/kg, a rather high dose) and cannulated rats as a model to evaluate the UWB effects on heart rate and blood pressure during exposure and immediately after exposure. The UWB exposure was usually less than 2 minutes. An earlier report was presented by Erwin and Hurt [1993] and details of the experiment were not given. However, the exposure was probably a 2 minute exposure by the "H-2" device which produced bipolar pulses at 250 kVlm (166 MW/m~ peak electric field intensity, 318-377 ps rise time, 5-10 os pulse duration, and 60 pps. Estimated whole-body average SAR was 20 mW/kg [Hurt 1999]. Jauchem et al. [1999] exposed rats to UWB pulses produced by a "Bournlea" pulse generator with pulses at a 19-21 kVlm peak electric field intensity, 327 ps rise time and 6 ns pulse duration. The exposure protocols were 2,000 pps for 0.5 s, 2,000 pps for 5 s, and 1,000 pps pulse train on for 2 s and off for 2 s for a total of 2 minutes. The estimated whole-body average SAR was 28 mW/kg [Hurt 1999]. None ofthese experiments included a sham-exposed group. Jauchem et al. [1998] also utilized the "Sandia" exposure system for rats. The experimental conditions were

248

sham, 50 pps (104 kVlm peak, 0.97 ns pulse duration, 177 ps rise time and 7.4 mWIkg estimated whole-body average SAR), 500 pps (102 kVlm peak, 0.97 ns pulse duration, 174 ps rise time and 71 mW/kg estimated whole-body average SAR) and 1,000 pps (87 kV/m peak, 0.99 ns pulse duration, 218 ps rise time and 114 mW/kg estimated whole-body average SAR). Results of these experiments revealed a stable heart rate and mean arterial pressure before, during and after UWB or sham exposures. However, the average heart rates (less than 265 beats per minute) of the Sprague-Dawley rats in these experiments were remarkably low for this species even if the cardiovascular stimulatory effect of the ketamine [Lippmann et al. 1983] is ignored. In conscious, free-roaming and cannulated rats, the average heart rate of Sprague-Dawley rats is between 332 and 364 beats per minute [Baron and Van Loon 1989, Ferrari et al. 1987, Houdi et al. 1995, Sparrow et al. 1987]. In addition, the delay time for manifestation of cardiovascular effects should be considered. Lu et al. [1992] demonstrated that it took 2 to 3 minutes for a heart rate change to appear in rats exposed to high peak power (7 MW/kg at the neck, 1 fJ-S pulse duration) and high average pulsed microwaves (110 W/kg at the neck) or high average CW microwaves (110 W /kg at the neck). In addition, the acute alteration in heart rate began to recover immediately after exposure.

In a non-invasive approach to evaluating the cardiovascular effects ofUWB exposure, Lu et al. [1999] measured the heart rate and arterial blood pressures (systolic, mean and diastolic) in conscious rats. Restraint adapted rats were exposed to sham, 500 Hz (93 kV/m, 180 ps rise time, 1 ns pulse duration, and 70 mWIkg estimated whole-body average SAR) and 1,000 Hz (85 kV/m, 200 ps rise time, 1.03 ns pulse width, and 121 mW/kg estimated whole-body average SAR) UWB pulses for 6 minutes. Cardiovascular endpoints were determined 3-4 days before exposure and 0.75 h, 24 h, 72 h, 1, 2, 3 and 4 weeks after exposure. Significant changes in heart rate were not found. On the other hand, significant decreases in systolic, mean, and diastolic pressures were observed in UWB exposed animals and the effect peaked at 2 weeks after UWB exposure. Recovery was still not complete by 4 weeks after exposure. In two subsequent, yet unpublished, experiments (Fig. 6), UWB exposures were sham, 125 Hz (91.3 kVlm peak, 156 ps rise time, 1.02 ns pulse duration, and 17 mW/kg estimated SAR), 250 Hz (94.0 kVlm peak, 168 ps rise time, 1.02 ns pulse duration, 38 mW/kg estimated SAR), 500 Hz (89.3 kV/m peak, 160 ps rise time, 0.91 ns pulse duration, and 59 mW/kg estimated SAR) and 1,000 Hz (85.4 kV/m, 167 ps rise time, 0.94 ns pulse duration and 112 mWIkg estimated SAR). The UWB-induced hypotension was robust (20 mm Hg), consistent, and persistent. The threshold appeared to be around 60 mW/kg estimated whole-body SAR.

In short, a response-SAR-exposure-duration relationship began to emerge from this limited number of studies on biological effects caused by UWB pulses. None of these average SARs can induce significant temperature increases. Additional characterization will be required to confirm and to elucidate response-SAR-exposure-duration relationships in other biological models. Dosimetry and mechanisms of action, either physical or biological, are still unresolved.

CONCLUSION

Analysis of biological effects and health implications of high peak power radio frequency pulses is a complicated endeavor by itself. The task is further complicated by the recent developments in methods of generating RF pulses by shifting from narrow-band pulses to carrierless pulses. Conventional radar is operated in a narrow band mode in which the fractional

249

DSHAM (15) ~125 Hz (5) ~250 Hz (5) II!!I!HSOO Hz (10) .1.000 Hz (10) 171.2 ,---,---,---....-----y-----r---.,.,-----,---r---,

166.2

~ 161.2

E .s 156.2

w ~ 151.2 (/J

f3 no n. ...J

146.2

« 141.2 ir w ~

!)t 136.2

U ::J 131.2 8 ~ (/J 126.2

121.2

0.14 1.48 1.97

... P< 0.05 by Dunnett's Test • p< 0.05 by AAOVA, r(4.40)

2.08 5.07. 4.48> 3.74. 4.18.

... ... ...

...

116.2 L-_-'-__ -'-__ -'-__ --'--__ -L __ -L __ --L __ --'-_--'

PRE 0.75 H 24 H 72 H 1 WI( 2 WK 3 WK 4WK

Figure 6. Development of hypotension in rats exposed to UWB pulses. Results of ANOV A at each sampling time point are shown at upper abscissa. Significant Fs are indicated by an asterisk. Significant differences between sham and exposed are shown by solid triangles.

bandwidth is less than 1% of the center (carrier) frequency. In carrierless RF pulses, the frequency domain of the pulse can span from DC to GHz which creates new challenges in dosimetry and in the potential mode of interactions between carrierless pulses and biological materials. The present review attempts to evaluate the progress in theoretical and experimental results of carrier and carrierless pulses.

A single RF pulse lasting from microseconds to seconds with adequate pulse energy is known to cause drastic acute biological effects. Examples in descending order of pulse energy are brain enzyme denaturation, stun and seizure, pain, decreased spontaneous activity and brain acetylcholine concentration, microwave induced whole-body movements, thermal sensation, startle modification and microwave hearing. The first four effects are associated with significant bulk: heating indicated by 2 ° C or more increase in body temperature. On the other hand, none

of the last four effects requires a significant bulk heating. They can be caused by a single pulse which is incapable of imparting adequate energy to cause an elevation in core temperature of animals. However, microwave induced whole-body movements are caused by a single pulse which results in 0.2 to 2 °C increase in subcutaneous temperature. On the other hand, a thermal sensation is elicited by partial body exposure in which less than 0.1 °C increase in cutaneous

temperature is noted. Startle modification and microwave hearing are produced by a pulse that is barely capable of transferring enough energy to cause less than a 10-3 or 1 ~ 0 C increase in tissue temperature of animals. The mechanism of action is not clear in startle modification and microwave induced whole-body movements. However, the mechanism of action is well established in the microwave auditory effect in which the cranial pressure wave created by thermoelastic expansion associated with an RF pulse is sensed by hair cells of the cochlear. The activation of thermoreceptors is undoubtedly involved in thermosensation. Both startle

250

modification and microwave induced whole body movements appears to be reflexive reactions which require receptor, reflex center and effectors. In other words, these last four effects are originated from activation of specialized nerve endings, mechanisms of which have not yet been adequately investigated.

The threshold ofa single pulse effect appears to have a critical duration. For an exposure period shorter than the critical duration, the threshold must be dependent on specific absorption and the threshold depends on specific absorption rate if the exposure duration is longer than the critical duration. Examples can be found in microwave hearing and microwave induced whole

body movements. However, the critical duration can vary with endpoint. This critical pulse duration appears to be around 30 j.lS for microwave hearing and 1 s for microwave induced

whole-body movements. The critical duration for a single pulse effect is not clear for other endpoints.

It has been proposed that high peak power RF pulses may have specific effects differing from those caused by a CW radiation. In addition, pulse modulation has been suspected to cause more enhancement of a given biological effect than the same one caused by CW radiation of equivalent average SAR. Microwave hearing can be considered to be specific to pulse modulation since it would not occur without the RF modulated pulse or pulses. On the other hand, pulse enhancement is not established reliably since many studies have failed to demonstrate its importance. Later studies aimed at comparing effectiveness of an induced biological effect between pulsed and CW RFs (duty factors > 104 or peak to average power ratio < 104) usually failed to confirm the earlier reports that indicated the existence of pulse enhancement. However,

studies have not been done to confirm or deny the enhancement effect of pulse modulation on injury of corneal endothelium, on brain cholinergic activity, and possibly on consolidation of working memory. A variety of behavioral effects have been observed in experiments employing "TEMPO" pulses. On the other hand, well-trained behaviors are resistant to alteration by "TEMPO" pulses. Due to the uniqueness of the "TEMPO" pulses (10-8 duty factor, 0.072 W/kg whole-body average SAR, noise, and x-ray) and the cost of maintaining such a system, it is highly recommended, by using the process of elimination, to experiment with 3.0 GHz CW at equivalent SARs to eliminate the existence of unsuspected low dose effects from CW microwaves.

A concrete conclusion regarding biological effects of carrierless RF is, at most, tentative because of the limited number of studies. The carrierless RF includes electromagnetic pulse (EMP) and ultra-wide-band (UWB) pulses. Although unconfirmed, a single EMP pulse with a

peak electric field of several hundred kVlm might interfere with the ability of animals to run a maze. Other than that report [Hirsh et at. 1968], most studies indicate no observable effect with EMP pulses with peak electric fields less than 100 kV/m. Studies of biological effects ofUWB pulses have begun to reveal the existence of UWB bioactivity at a peak electric field intensity around 100 kVlm. Because of health and medical implications, UWB-induced hypotension needs to be independently confirmed and the dose-response characteristics further explored.

Acknowledgment: The authors are in part supported by the US. Army Medical Research and Materiel Command under contract No. DAMD-17-94-C-4069 with McKessonHBOC BioServices. The views, opinions and/or findings contained in this report are those of the authors and should not be construed as an offiCial Department of the Army position, policy or decision unless so designated by other documentation. Approvedfor public release, distribution is unlimited

251

REFERENCES

Adair, RK., 1991, Biological effects on the cellular level of electric field pulses, Health Phys. 61: 395-399.

Adair, RK., 1995, Ultrashort microwave signals: a didactic discussion, Aviat. Space Environ. Med. 66: 792-794.

Agee, F.I, Schlfield, D.W., Prather W., and Burger, J.W., 1995, Powerful ultra-wide band RF emitters: status and challenges, In: Intense Microwave Pulses III, H.E. Brandt, ed., Proceeding of the Int. Soc. Optical Eng. vol 2557, pp 98-109.

Agouridis, D.C., and Easterly, C.E., 1989, EMP simulators and public safety: an analysis, Bioelctromagnetics 10: 355-360.

Akyel, Y., Raslear, T.G., and Serfini, R, 1990, Exposure of rodents in a parallel plate EMP simulator, Part II: behavioral effects, Proc. Ann. Conf. IEEE EMBS 12: 1551-1152.

Akyel, Y., Raslear, T.G., Bates, F., and DeAngelis, M., 1992, Behavioral responses of rats to acute EMP fields, Proc. Ann. Conf. IEEE EMBS 14: 287-288.

Akyel, Y., and Raslear, T.G., 1993, The effects of acute EMP fields on the behavior of rats, In: EMP Human Health Effects Science Review Panel Proceedings, J. de Lorge and W. Mick, eds., U.S. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 45-57.

Akyel, Y., Akyel, B., and Raslear, T.G., 1993a, Further characterization of microwaveevoked whole-body movement. Proc. Ann. Conf. IEEE EMBSI5: 1451-1452.

Akyel, Y., Belt, M., Raslear, T.G., and Hammer, RM., 1993b, The effects of high-peak power pulsed microwaves on treadmill performance in the rat, In: Electricity and Magnetism in Biology and Medicine, M. Blank, ed., San Francisco Press, San Francisco, CA, pp 668-670.

Albanese, R, Blaschak, J., Medina, R., and Penn I, 1994, Ultrashort electromagnetic signals: biophysical questions, safety issues, and medical opportunities, Aviat. Space Environ. Med. 65lSupplement: AI16-AI20.

Albert, E.N., 1979, Reversibility of microwave-induced blood-brain-barrier permeability, Radio Sci. 14/6 Supplement: 323-327.

Arber, S.L., 1981, The effect of microwave radiation on passive membrane properties of snail neurons, I Microwave Power 16: 15-20.

Arber, S.L., and Lin, IC., 1985a, Microwave-induced changes in nerve cells: effects of modulation and temperature, Bioelectromagnetics 6: 257-270.

Arber, S.L., and Lin, IC., 1985b, Extracellular calcium and microwave enhancement of membrane conductance in snail neurons, Radiat. Environ. Biophys. 24: 149-156.

Baranski, S., 1971, Effect of chronic microwave irradiation on the blood forming system of guinea pigs and rabbits, Aerospace Med. 42: 1196-1199.

Baranski, S., 1972a, Histological and histochemical effect of microwave irradiation on the central nervous system of rabbits and guinea pig, Am. I Phys. Med. 51: 182-191.

Baranski, S., 1972b, Effect of microwaves on the reactions of white blood cells system, Acta Physiol. Pol. 23: 687-695.

Baranski, S., and Czerski, P., 1976, Microwave effects on blood and the blood-forming system, In: BiolOgical Effects of Microwaves, Dowden, Hutchinson & Ross, Inc., Stroudsburg, PA, pp 137-146.

252

Barron, B.A, and Van Loon, GR, 1989, Role of sympathoadrenomedullary system in cardiovascular response to stress in rats, J. Autonomic Nervous System 28: 179-188.

BaUIn, S.J., 1979, Tests of biological integrity in dogs exposed to an electromagnetic pulse environment, HealthPhys. 36: 159-165.

Baum, S.J., Ekstrom, M.E., Skidmore, W.O., Wyant, D.E., and Atkinson, J.L., 1976, Biological measurements in rodents exposed continuously throughout their adult life to pulsed electromagnetic radiation, Health Phys. 30: 161-166.

Bernardi, P., and D'Inzeo, G., 1984, A nonlinear analysis of the effects of transient electromagnetic fields on excitable membrane, IEEE Trans. Microwave Theory Tech. 32: 670-679.

Birenbaum, L., Grosof, G.M., Rosenthal, S.W., and Zaret, M.M., 1969, Effect of microwaves on the eye, IEEE Trans. Biomed. Eng. 16: 7-14.

Birenbaum, L., Kaplan, I.T., Metlay, W., Rosenthal, S.W., and Zaret, M.M., 1975, Microwave and Infra-red effects on heart rate, respiration rate and subcutaneous temperature of the rabbit, J. Microwave Power 10: 3-18.

Blair, H.A., 1932, On the intensity-time relations for stimulation by electric current, J. Gen. Physiol. 15:177-185,709-729,731-755.

Blick, D.W., Adair, E.R, Hurt, W.O., Sherry, C.J., Walters, T.J., and Merritt, J.H., 1997, Thresholds of microwave-evoked warmth sensations in human skin, Bioelectromagnetics 18: 403-409.

Bock, T.J., 1988, FMP test of implantable cardiac pacemakers, U.S. Army Laboratory Command, Harry Diamond Laboratories, Adelphi, MD, Report HDL-TR-2197.

Brown, D.O., Lu, S.-T., and Elson, B.C., 1994, Characteristics of microwave evoked body movements in mice, Bioelectromagnetics 15: 143-161.

Brown, P.VK., and Wyeth, N.C., 1983, Laser interferometer for measuring microwaveinduced motion in eye lens in vitro, Rev. Sci. Instrum. 54: 85-89.

Cain, C.A., and Rissman, W.J., 1978, Mammalian auditory responses to 3.0 GHz microwave pulses, IEEE Trans. Biomed. Eng. 25: 288-293.

Chatterjee, 1.0., Wu, D., and Gandhi, O.P., 1986, Human body impedance and threshold currents for perception and pain for contact hazard analysis in the VLF-MF band, IEEE Trans. Biomed. Eng. 33: 486-494.

Chen, J.-Y., and Gandhi, O.P., 1991, Currents induced in an anatomically based model of human for exposure to vertically polarized electromagnetic pulses, IEEE Trans. Microwave Theory Tech. 30: 31-39.

Chou, C.-K., and Guy, AW., 1978, Effects of electromagnetic fields on isolated nerve and muscle preparations, IEEE Trans. Microwave Theory Tech. 26: 141-147.

Chou, C.K., Guy, AW., Galambos, R, and Lovely, RH., 1975, Cochlea microphonics generated by microwave pulses, J. Microwave Power 10: 361-367.

Chou, C.K., Guy, AW., and Galambos, R, 1976, Microwave-induced auditory response: cochlear microphonics, In: Biological Effects of Electromagnetic Waves, Vol. I, C.C. Johnson and M.L. Shore, eds., FDAlBRH, Rockville, MD., HEW Publication (FDA) 77-8010, pp 291-296.

Chou, C.K., Guy, AW., and Galambos, R, 1982a, Auditory perception of radio-frequency electromagnetic fields. J. Acoust. Soc. Am. 71: 1321-1334.

Chou, C.-K., Guy, AW., McDougall, J.A. and Han, L.-F., 1982b, Effects of continuous and pulsed chronic microwave exposure on rabbits, Radio Sci. 17/5 Supplement: 185-193.

253

Chou, C.K., Han, L.F., and Guy, AW., 1980, Microwave radiation and heart-beat rate of rabbits, l Microwave Power 15: 87-93.

Clapman, R.M., and Cain, C.A, 1975, Absence of heart rate effects in isolated frog heart irradiated with pulse modulated microwave energy, l Microwave Power 10: 411-419.

Cleary, S.F., Nickless, F., Liu, L.M., and Hoffinan, R, 1980, Studies of exposure of rabbits to electromagnetic pulsed fields, Bioelectromagnetics 1: 345-352.

Cordts, RE., Merritt, lB., Erwin, D.N., Hardy, K.A, and Yochmowitz, M.G., 1988, Behavioral Response of Rats Exposed to High-Power Microwave Radiation, USAF School of Aerospace Medicine, Brooks Air Force Base, TX, USAF SAM-TR-87-30 (DTIC ADAI92199).

Creighton, M.O., Larsen, L.E., Stewart-DeHaan, PJ., Jacobi, lH., Sanwal, M., Baskerville, lC., Bassen, H.E., Brown, D.O., and Trevithick, lR., 1987, In vitro studies of microwave-induced cataract, II. comparison of damage observed for continuous wave and pulsed microwaves. Exp. Eye Res. 45: 233-246.

Czerska, E.M., Elson, E.C., Davis, C.C., Swicord, M.L., and Czerski, P., 1992, Effects of continuous and pulsed 2450-MHz radiation on spontaneous lymphoblastoid transformation of human lymphocytes in vitro, Bioelectromagnetics 13: 247-259.

Czerski, P., Paprocka-Stonka, E., Siekierzynski, M., and Stolarska, A, 1974, Influence of microwave radiation on the hematopoietic system, In: Biological Effects and Health Hazards of Microwave Radiation, P. Czerski, Ch. Silverman, K. Ostrowski, M.l Suess, M.L. Shore, and B. Waldeskog, eds., Polish Medical Publishers, Warsaw, Poland, pp 67-74.

Dalziel, C.F., and Mansfield, T.H., 1950, Effects of frequency on perception currents, AlEE Trans. 69: 1162-1168.

Davis, M., 1980, Neurochemical modulation of sensory-motor reactivity: acoustic and tactil startle reflexes, Neurosci. Beh. Rev. 4: 241-263.

D'Andrea, lA, Knepton, l, Cobb, B.L., Klauenberg, B., and Merritt, lH., 1989, High Peak Power Microwave Pulses at 2.37 GH: no effects on Vigilance performance in

monkeys, Naval Aerospace Medical Research Laboratory, Pensacola, FL, and USAF School of Aerospace Medicine, Brooks AFB, TX, NAMRL-1348, and USAFSAM TR-89-21 (DTIC ADA219570).

D'Andrea, lA, Cobb, B.L., Knepton, l, Shull, RN., and Klauenberg, B.l, 1990, No Effects of High-Peak-Power Microwave Pulses at 2.36 GHz on Behavioral Performance in Monkeys, Naval Aerospce Medical Research Laboratory, Pensacola, FL, and USAF School of Aerospace Medicne, Brooks AFB, TX, NMRL-1358 and USAF SAM TR-90-14 (DTIC ADA283667).

D'Andrea, lA, Cobb, B.L., Knepton, l, and Bates, F., 1993, Behavioral Performance in Monkeys Exposed to TEMPO High-Peak-Power Microwave Pulses at 3 GHz, Naval Medical Research Laboratory, Pensacola, FL, NAMRL-1389, (DTIC ADA280551).

D'Andrea, lA, Thomas, A, and Hatcher, D.l, 1994, Rhesus monkey behavior during exposure to high-peak-power 5.62-GHz microwave pulses, Bioelectromagnetics 15: 163-176.

deLorge, lO., 1984, Operant behavior and colonic temperature ofMacaca mulatta exposed to radiofrequency fields at and above resonant frequency, Bioelectromagnetics 5: 233-246.

254

Eijkman, E., and Vendrik, AlH., 1961, Dynamic behavior of the wann sense organ, J. Exp. Psychol. 62: 403-408.

Ellis, V.J., 1991, The effects of electromagnetic pulse (EMP) on cardiac pacemakers, U.S. Army Laboratory Command, Harry Diamond Laboratories, Adelphi, MD, Report HDL-TR-2197 (DTIC ADA242990).

Erwin, D.N., and Hurt, W., Biological effects of ultra wide band emmissions, In: EMP Human Health Effects Science Review Panel Proceedings, J. de Lorge and W. Mick, eds., u.s. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 91-95.

Ferrari, AU., Daffonchio, A, Albergati, F., and Mancia, G., 1987, Inverse relationship between heart rate and blood pressure variability in rats, Hypertension 10: 533-537.

FleshIer, M., 1965, Adequate acoustic stimulus for startle reaction in the rat, J. Com. Physiol. Psychol. 60: 200-207.

Foster, K.R., and Finch, E.D., 1974, Microwave hearing: evidence for thermoacoustical auditory stimulation by pulsed microwave. Science 185: 246-248.

Foster, P.R., 1995, Special antenna requirements for UWB Radar Systems, Introduction, In: Introduction to Ultra-Wideband Radar System, lD. Taylor, ed., CRC Press, Boca Raton, FL, pp 107-149.

Frei, M.R., and Jauchem, lR., 1989, Effects of2.8-GHz Microwaves on restrained and ketamine-anesthetized rats. Radiat. Environ. Biophys. 28: 155-164.

Frei, M., Jauchem, J., and Heinmets, F., 1988, Physiological effects of2.8 GHz radiofrequency radiation: a comparison of pulsed and continuous-wave radiation, J. Microwave Power 23: 85-93.

Frei, M., Jauchem, J., and Heinmets, F., 1989, Thermoregulatory responses of rats exposed to 9.3-GHz radiofrequency radiation, Radiat. Environ. Biophys. 28: 67-77.

Frei, M.R., Jauchem, J.R., and Padilla, J.M., 1989a, Thermal and physiological changes in rats exposed to CW and pulsed 2.8 GHz radiofrequency radiation in E and H orientation, Int. J. Radiat. Biol. 56: 1033-1044.

Frey, AH., 1961, Auditory system response to radio frequency energy, Aerospace Med. 32: 1140-1142.

Frey, AH., and Feld, S.R., 1975, Avoidance by rats of illumination with low power nonionizing electromagnetic energy, J. Compo Physiol. Psychol. 89: 183-188.

Frey, AH., Feld, S.R., and Frey, B., 1975, Neural function and behavior: defining the relationship. Ann. N.Y. Acad. Sci. 247: 433-439.

Frey AH., and Seifert, E., 1968, Pulse modulated UHF energy illumination of the heart associated with change in heart rate. Life Sci. 7: 505-512.

Fuerer, D., 1991, Electronic countermeasures in wideband spectrum systems, In: Propagation Limitations for Systems Using Band Spreading, Advisory Group for Aerospace Research and Development, AGARD-LS-I72, pp 1-11.

Gailey, P.C., and Easterly, C.E., 1993, Cell membrane potentials induced during exposure to EMP fields, In: EMP Human Health Effects Science Review Panel Proceedings, J. de Lorge and W. Mick, eds., U.S. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 133-139.

Gao, B.-Q., Gandhi, O.P., Mathur, S., Bates, F., and Bassen, H., 1992, Currents induced in a model of a rat exposed to EMP in a parallel-plate line: calculations and experimental results, Bioelectromagnetics 13: 439-443.

255

Geddes, L.A., and Baker, L.E., 1989, Principles of Applied Biomedical Instrumentation, 3rd edition, John Wiley, New York, NY, p. 961.

Ginsburg, KS., Lin, I.C., and 0' Neill, W.D., 1992, Microwave effects on input resistance and action potential firing of snail neuron, IEEE Trans. Biomed. Eng. 39: 1011-1021.

Goldman, H., Lin, I.C., Murphy, S., and Lin, M.F., 1984, Cerebrovascular permeability to B~b in the rat after'exposure to pulsed microwaves, Bioelectromagnetics 5: 323-330.

Gruenau, S.P., Oscar, KJ., Folker, M.T., and Rapoport, SJ., 1982, Absence of microwave effect on blood-brain barrier permeability to 14C-sucrose in the conscious rat, Exp. Neurobiol. 75: 299-307.

Guy, A.W., 1989, Analysis ofEMP induced currents in human body by NEC method of moments, Proc. Ann. Int. Conf IEEE EMBS 11: 1132-1133.

Guy, A. W., 1990, Maximum induced currents, SAR and absorbed energy density in human and non-human primates exposed to EMP, Proc. Ann. Int. Conf IEEE EMBS 12: 1547-1548.

Guy, A.W., and Chou, C.-K, 1982, Effects of high-intensity microwave pulse exposure of rat brain, Radio Sci. 17/5 Supplement: 169-178.

Guy, W., Chou, C.K, Lin, I.C., and Christense, D., 1975, Microwave-induced acoustic effects in mammalian auditory systems and physical materials, N.Y. Acad. Sci. 247: 194-218.

Hagmann, M.J., 1993, Human dosimetry in EMP exposures with minimally-perturbing current probes, In: FMP Human Health Effects Science Review Panel Proceedings, J. de Lorge and W. Mick, eds., U.S. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 59-78.

Hamrick, P., and McRee, D.I., 1980, The effect of2450 MHz microwave irradiaiton on the heart rate of embryonic quail, Health Phys. 38: 261-268.

Hanson-Mild, K, Lovdahl, L., Lovstrand, K-G., and Lovtrup, S., 1982, Effect of highvoltage pulses on the viability of human leukocytes in vitro, Bioelectromagnetics 3: 213-218.

Hendler, E., 1968, Cutaneous receptor response to microwave irradiation, In: Thermal , Problems in Aerospace Medicine, I.D. Hardy, ed., Unwin Ltd., Surry, pp 149-161.

Hendler, E., Hardy, I.D., and Murgatroyd, D., 1963, Skin heating and temperature sensation produced by microwave irradiaiton, In: Temperature: Its Measurement and

Control in Science and Industry, C.M. Herzfeld, ed., Reinhold, New York, NY, pp 211-230.

Hirsh, F.G., McGiboney, D.R., and Harnish, T.D., 1968, The psychologic consequences of exposure to high density pulsed electromagnetic energy, Int. I. Biometeror. 12: 263-270.

HjereSen, D.L., Doctor, S.R, and Sheldon, RL., 1978, Shuttle side preference as mediated by pulsed microwave and conventional auditory cues, In: Electromagnetic Fields in BiolOgical Systems, S.S. Stuchly, ed., IMPI, Edmonton, Canada, pp 194-214.

Houdi, A.A., Dowell, RT., and Diana, I.N., 1995, Cardiovascular responses to cigarette smoke exposure in restrained conscious rats, I. Pharmacol. Exp. Therap. 275: 646-,653.

Hurt, W.D., 1999, Dosimetry in animals exposed to ultra-wide-band pulses produced by various Air Force devices, personal communication.

256

ICNIRP 1998, Guidelines for limiting exposure to time-varying electric, magnetic, and electromagnetic fields (up to 300 GHz), Health Phys. 74: 494-522.

IEEE, 1999, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz, Institute of Electrical and Electronics Engineers, New York, NY.

Jauchem, lR, Seaman, RL., Lehnert, HM., Mathur, S.P., Ryan, KL., Fre~ M.R., and Hurt, W.D., 1998, Ultra-wideband electromagnetic pulses: lack of effects on heart rate and blood pressure during two-minute exposures of rats, Bioelectromagnetics 19: 330-333.

Jauchem, IR, Frei, M.R, Ryan, KL., Merritt, IH., and Murphy, MR, 1999, Lack of effects on heart rate and blood pressure in ketamine-anesthetized rats briefly exposed to ultra-wideband electromagnetic pulses, IEEE Trans. Biomed. Eng. 46: 117-120.

Johnson, R.B., Meyers, D.E., Guy, AW., Lovely, RH, and Galambos, R, 1976, Discrimininative control of appetitive behavior by pulsed microwave radiation in rats, In: Biological Effects of Electromagnetic Waves, C.C. Johnson, and M.L. Shore, eds., FDA, Rockville, MD, HEW Publication (FDA) 77-8010, pp 238-247.

Jokela, K, 1997, Electrophysiological considerations relevant to the limiting of pulsed electric and magnetic fields, Health Phys. 72: 408-417.

Justesen DR, Adair, E.R, Stevens, IC., Bruce-Wolfe, V., 1982, A comparative study of human sensory thresholds: 2450-MHz microwaves vs far-infrared radiation, Bioelectromagnetics 3: 117-125.

Kanter, G.S., 1960, Glomerular filtration and renal plasma flow during hyperthermia, Am. J. Physiol. 198: 1044-1048.

Klauenberg, B.I, Merritt, IH., and Erwin, D.N., 1988, Behavioral Effects of Exposure to

the TEMPO High-Power Microwave System, USAF School of Aerospace Medicine Brooks Air Force Base, TX, USAF SAM-TR-87-35 (OTIC ADAI93305).

Kues, H.A, Hirst, L.W., Lutty, G.A., D'Anna, S.A, and Dunkelberger, G.R., 1985, Effects of2.45 GHz microwaves on primate corneal endothelium, Bioelectromagnetics 6: 177-188.

Lacourse, J.R, Vogt, M.C., Miller, M.T., and Selikowitz, S.M., 1985, Effect of highfrequency current on nerve and muscle tissue, IEEE Trans. Biomed. Eng. 32: 82-86.

Lai, H., 1994, Neurological effects of radio frequency electromagnetic radiation, In: Advances in Electromagnetic Fields in Living Systems, IC. Lin ed., Plenum Press, New York, NY, pp. 27-80.

Lai, H., Horita, A, Chou, C.K., and Guy, A W., 1987, Low-level microwave irradiaiton affects central cholinergic activity in the rat, J. Biochem. 48: 40-45.

Lai, H, Horita, A, and Guy, AW., 1988, Acute low-level microwave exposure and central cholinergic activity: studies on irradiation parameters, Bioelectromagnetics 9: 355-362.

Lai, H., Horita, A, and Guy, AW., 1994, Microwave irradiation affects radial-ann maze performance in the rat, Bioelectromagnetics 15: 95-104.

Lebovitz, RM., 1983, Pulse modulated and continuous wave microwave radiation yield equivalent changes in operant behavior of rodents, Physiol. Beh. 30: 891-898.

Lebovitz, RM., and Seaman, RL., 1977, Microwave hearing: the response of single auditory neurons in cat to pulsed microwave radiation, Radio Sci. 12/6 Supplement: 229-236.

257

Lee, R.C., Gaylor, D.C., Bhatt, D., Israel, D.A., 1988, Role of cell membrane rupture in the pathogenesis of electrical trauma, l Surg. Res. 44: 709-719.

Lenox, R.H., Meyerhoff, lL., Gandhi, O.P., and Wray, H.L., 1977, Regional levels of cyclic AMP in rat brain; pitfalls of microwave interaction. J. Clyclic Nucleotide Res. 3: 367-379.

Lin, J.C., 1975, Interaction of electromagnetic transient radiation with biological materials, IEEE Trans. Electromagnetic Compatibility 17: 93-97.

Lin, J.C., 1976, Electromagnetic pulse interaction with mammalian cranial structures, IEEE Trans. Biomed. Eng. 23: 61-65.

Lin, lC., 1976a, Microwave auditory effect - a comparison of some possible transduction mechanisms, J. Microwave Power 11: 77-81.

Lin, lC., 1976b, Microwave induced hearing sensation: some preliminary theoretical observations, l Microwave Power 11: 295-298.

Lin, lC., 1976c, Theoretical analysis of microwave-generated auditory effects in animals and man, In: Biological Effects of Electromagnetic Waves, C.C. Johnson and M.L. Shore, eds., FDAlBRH, Rockville, MD, HEW Publication (FDA) 77-8010, pp 36-48.

Lin, J.C., 1977, On microwave-induced hearing sensation, IEEE Trans. Microwave Theory Tech., 25: 605-613.

Lin, l C., 1978, Microwave Auditory Effects and Applications, Charles C. Thomas, Springfield, IL.

Lin, lC., 1980, Microwave auditory phenomenon, Proc. IEEE 68: 67-73. Lin, lC., 1989, Pulsed radiofrequency field effects in biological systems, In:

Electromagnetic Interaction with Biological Systems," J.C. Lin ed., Plenum Press, New York, NY, pp 165-177.

Lin, J.C., 1990, Auditory perception of pulsed microwave radiation, In: Biological Effects and Medical Applications of Electromagnetic Energy, O.P. Gandhi, ed., Prentice Hall, Englewood Cliffs, NJ, pp 277-318.

Lin, lC., and Lam, C.K., 1976, Coupling of Gaussian electromagnetic pulse into a musclebone model of biological structure, I Microwave Power 11: 67-75.

Lin, J.C., and Lin, M.F., 1980, Studies on microwave and blood-brain barrier interaction, Bioelectromagnetics 1: 313-323.

Lin, IC., and Lin, M.F., 1982, Microwave hyperthermia-induced blood-brain barrier alteration, Radiat. Res. 89: 77-87

Lin, J.C., Meltzer, R.I, and Redding, F.K., 1979, Microwave-evoked brainstem potentials in cats, I Microwave Power 14: 291-296.

Lin, J.e., Meltzer, R.I, and Redding, F.K., 1982, Microwave-evoked brainstem auditory responses, Proc. San Diego Biomed. Symp. 17: 461-465.

Lin, J.C., Su, I-L., and Wang, Y., 1988, Microwave-induced thermoelastic pressure propagation in the cat brain, Bioelectromagnetics 9: 141-147.

Lin, J.C., Wu, C.-L., and Lam, C.K., 1975, Transmission of electromagnetic pulse into the head, Proc. IEEE 63: 1726-1727.

Lippmann, M., Appel, P.L., Mok, M.S., and Shoemaker, W.C., 1983, Sequential cardiorespiratory patterns of anesthetic induction with ketamine in critically ill patients, Critical Care Med. 11: 730-733.

Liu, L.M., Rosenbaum, F.1., and Pickard, W.F., 1976, The insensitivity offrog heart rate to pulse modulated microwave energy, 1. Microwave Power 11: 225-232.

258

Lu, S.-T., Brown, D.O., Johnson, C.E., Mathur, S.P., and Elson, E.C., 1992, Abnonnal cardiovascular responses induced by localized high power microwave exposure, IEEE Trans. Biom. Eng. 39: 484-492.

Lu, S.-T., Mathur, S.P., Akyel, Y, and Lee, l, 1999, Ultrawide-band electromagnetic pulses induced hypotension in rats, Physiol. Beh. 65: 753-761.

Lu, S.-T., Roberts, N.l, Jr., and Michaelson, S.M., 1983, A dual vial waveguide exposure facility for examining microwave effect in vitro, 1 Microwave Power 18: 121-131.

Mathur, S., Bates, F., and Bassen, H., 1990, Exposure of rodents in a parallel plate EMP simulator, Part I: Engineering and dosimetry, Proc. Ann. Int. Com. IEEE EMBS 12: 1549-1550.

Mattsson, lL., and Oliva, S.A, 1976, Effect of electromagnetic pulse on avoidance behavior and electroencephalogram of a rhesus monkey, Aviat. Space Environ. Med. 47: 644-648.

McRee, OJ., and Wachtel, H., 1980, The effects of microwave radiation on the vitality of isolated frog sciatic nerves, Radiat. Res. 82: 536-546.

McRee, DJ., and Wachtel, H., 1982, Pulse microwave effects on nerve vitality, Radiat. Res. 91: 212-218.

McLees, B.D., Finch, E.D., and Albright, M.L., 1972, An examination of regenerating hepatic tissue subjected to radio-frequency irradiation, 1 Appl. Physiol. 32: 78-85.

Merritt, lH., Chamness, AF., and Allen, S.J., 1978, Studies on blood-brain barrier penneability after microwave-radiation, Radiat. EnViron. Biophy. 15: 367-377.

Merritt, lH., Kiel, lL., and Hurt, W.O., 1995, Considerations for human exposure standards for fast-rise-time high-peak-power electromagnetic pulses, Aviat. Space Environ. Med. 66: 586-589.

Michaelson, S.M., and Lin, lC., 1987, Personnel protection, protection guides, and standards, In: Biological Effects and Health Implications of Radiofrequency Radiation, ch. 20, Plenum Press, New York, NY, pp 637-658.

Michaelson, S.M., and Lin, lC., 1987a, The common integument (skin), In: Biological Effects and Health Implications of Radiofrequency Radiation, ch. 17, Plenum Press, New York, NY, pp 539-558.

Miller, SA, Bronson, M.E., and Murphy, M.R., 1999, Ultrawideband radiation and pentylenetetrazol-induced convulsions in rats, Bioelectromagnetics 20: 327-329.

Modak, AT., Stavinoha, W.B., and Deam, AP., 1981, Effect of short electromagnetic pulses on brain acetylcholine content and spontaneous motor activity of mice, Bioelectromagnetics 2: 89-92.

Moore, F.R., and Simons, T., 1993, Orientation behavior of migratory birds in response to EMPRESS II-simulated EMP, In: EMP Human Health Effects Science Review Panel Proceedings, 1 de Lorge and W. Mick, eds., U.S. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 97-112.

Mouneimne, Y, Tosi, P.-F., and Nicolau, C., 1993, Electroinsertion of proteins into membranes: a novel approach to the study of membrane receptors, In: Electricity and Magnetism in Biology and Medicine, M. Blank, ed., San Francisco Press, San Francisco, CA, pp 112-114.

Muhm, J.M., 1992, Mortality investigation of workers in an electromagnetic pulse test program, 1 Occup. Med. 34: 287-292.

259

Neilly, J.P., and Lin, I.C., 1986, Interaction of ethanol and microwaves on the blood-brain barrier of rats. Bioelectromagnetics 7: 405-414.

Neshev, N.N., and Kirilova, E.I., 1996, Environmental-health aspects of pulse-modulated microwaves, Rev. Environ. Health 11: 85-88.

Neumann, E., Werner, E., Sprafke, A, and Kruger, K., 1993, Membrane electroporation: two are a molecular mechanism, In: Electricity and Magnetism in Biology and Medicine, M. Blank, ed., San Francisco Press, San Francisco, CA, pp 109-111.

Newman, E., and Katchalsky, A, 1972, Long-lived conformation changes induced by electric impulses in biopolymers, Proc. Nat. Acad. Sci. 69: 993-997.

NCRP Report 86, 1986, Perception ofPFEM fields, In: Biological Effects and Exposure Criteria for Radiofrequency Fields, National Research Council on Radiation Protection and Measurements, Bethesda, MD, pp 175-180.

Oldendorf, W.H., 1970, Measurement of brain uptake of radio labeled substances using tritiated water internal standard, Brain Res. 24: 372-376.

Olsen, R.G., and Lin, I.C., 1981, Microwave pulse-induced acoustic resonances in spherical head models, IEEE Trans. Microwave Theory Tech. 29: 1114-1117.

Olsen, R.G., and Lin, I.C., 1983, Microwave-induced pressure waves in mammalian brains, IEEE Trans. Biomed. Eng. 30: 289-294.

Oscar, KJ., Gruenau, S.P., Folker, M.T., and Rapoport, S.I., 1981, Local cerebral blood flow after microwave exposure, Brain Res. 204: 220-225.

Oscar, KI., and Hawkins, T.D., 1977, Microwave alteration of the blood-brain barrier system of rats, Brain Res. 126: 281-293.

Paile, W., Jokela, K, Koivistoinen, S., and Salomaa, S., 1995, Effect of 50 Hz magnetic fields and spark discharges on human lymphocytes in vitro, Bioelectrochem. Bioenergy 36: 15-22.

Pakhomov, AG., Dubovick, B.V., Degtyariov, I.G., and Pronkevick, AN., 1995, Microwave influence on the isolated heart function: I. effect of modulation, Bioelectromagnetics 16: 241-249.

Pakhomova, O.N., Belt, M.L., Mathur, S.P., Lee, J.C., and Akyel, Y, 1997, Lack of genetic effects of ultra wide-band electromagnetic radiation in yeast, Electro-Magnetobiol. 16: 195-201.

Pakhomova, O.N., Belt, M.L., Mathur, S.P., Lee, I.C., and Akyel, Y, 1998, Ultra-wide band electromagnetic radiation does not affect UV -induced recombination and mutagenesis in yeast, Bioelectromagnetics 19: 128-130.

Patrick, E.L., [1993], Human health issues of woodbridge research facility's electromagnetic pulse simulators, In: EMP Human Health Effects Science Review Panel Proceedings, J. de Lorge and W. Mick, eds., U.S. Navy Theater Nuclear Warfare Program and Operational Medicine and Fleet Support, Washington, D.C., pp 23-37.

Porsolt, R.D., Anton, G., Blavet, N., and Jalfre, M., 1978, Behavioral despair in rats: a new model sensitive to antidepressant treatments, Eur. 1. Pharmacol. 47: 379-391.

Postow, E., and Swicord, M.L., 1996, Modulated fields and "window" effects, In: CRC Handbook of Biological Effects of Electromagnetic Fields, C. Polk and E. Po stow, eds., CRC Press, Boca Raton, FL, pp 535-580.

Presman, AS., and Levitina, N.A., 1963, Nonthermal action of microwaves on the rhythm of cardiac contractions in animals, Bull. Exp. BioI. Med. 53: 154-157.

260

Preston, E., Vavasour, E.J, and Assenheim, H.M., 1979, Permeability ofthe blood-brain barrier to mannitol in the rat following 2450 MHz microwave irradiation, Brain Res.

174: 109-117. Rafferty, C.N., and Knutson, JR., 1987, Effects of pulsed microwave fields on soluble

proteins and liposomes, Proc. Ann. Int. Conf.IEEE EMBS 9: 701-702. Raslear, T.G., Akyel, Y, Bates, F., Lu, S.-T., and Elson, E.C., 1993a, The Behavioral

Toxicology of High-Peak, Low Average Power, Pulsed Microwave Irradiation, USAMRDC, Washington, D.C. (DTIC ADA268392).

Raslear, T.G., Akyel, Y, Bates, F., Belt, M., and Lu, S.-T., 1993b, Temporal bisection in rats: The effects of high-peak-power pulsed microwave irradiation, Bioelectromagnetics 14: 459-478.

Raslear, T.G., Akyel, Y, Lu, S.-T., Swearengen, JR, Varle, D., DeAngelis, M.L., and Seaman, RL., 1992, CW microwave fields evoke body movements in bilaterally cochleotomized rats, Proc. Ann. Int. Conf. IEEE EMBS 14: 295-296.

Raslear, T.G., Akyel, Y., Serafini, R, Bates, F., and Belt, M., 1991a, Memory consolidation

in the rat following high-peak power pulsed microwave irradiation. Proc. Ann. Int. Conf. IEEE EMBS 13: 958-959.

Raslear, T.G., Akyel, Y, Serafini, R, Bates, F., and Belt, M., 1991b, Food demand and circadian rhythmicity following high-peak power pulsed microwave irradiation. Proc. Ann.Int. Conf.IEEE EMBS 13: 962-963.

Reilly, JP., 1991, Magnetic field excitaton of peripheral nerves and the heart: a comparison of thresholds, Med. BioI. Eng. Comput. 29: 571-579.

Reilly, J.P., 1998, Maximum pulsed electromagnetic field limits based on peripheral nerve stimulation: Application to IEEE/ANSI C95.1 electromagnitic field standards, IEEE Trans. Biomed. Eng. 45: 137-141.

Riu, P.J, Foster, K.R., Blick, D.W., and Adair, E.R.,1997, A thermal model for human threshold of microwave-evoked warmth sensations, Bioelectromagnetics 18: 578-583.

Roberti, B., Heebels, G.H., Hendricx, JC.M., de Greef, A.H.A.M., and Wolthuis, O.L., 1975, Preliminary investigations of the effects of low-level microwave radiation on spontaneous motor activity in rats. Ann. N.Y Acad. Sci., 247: 416-424.

Roberts, N.J, Jr., Lu, S.-T., and Michaelson, S.M., 1983, Human leukocyte functions and the U. S. Safety standard for exposure to radio-frequency radiation, Science 220: 318-320.

Roberts, N.J., Jr., Lu, S.-T., and Michaelson, S.M., 1985, Hyperthermia and human leukocyte functions: DNA, RNA, and total protein synthesis after exposure to <41 0 or

>42.5 0 hyperthermia, Cancer Res. 45: 3076-3082.

Roberts, N.J., Jr., Michaelson, S.M., and Lu, S.-T., 1984, Exposure of human mononuclear leukocytes to microwave energy pulse modulated at 16 and 60 Hz, IEEE Trans.

Microwave Theory Tech. 32: 803-807. Roberts, N.J., Jr., Michaelson, S.M., and Lu, S.-T., 1987, Mitogen responsiveness after

exposure of influenza virus-infected human mononuclear leukocytes to continuous or pulse-modulated radiofrequency radiation, Radiat. Res. 110: 353-361.

Sanders, A.P., Joines, W.T., and Allis, JW., 1985, Effects of continuous-wave, pulsed, and sinusoidal-amplitude-modulated microwaves on brain energy metabolism, Bioelectromagnetics 6: 89-97.

261

Sandler, S.S., Smith, G.S., and Albert, E.N., 1975, Electromagnetic field effects in nerve tissue, Aviat. Space Environ. Med. 46: 1414-1417.

Schwartz, D.C., and Koval, M., 1989, Conformational dynamics of individual DNA molecules during gel electrophoresis, Nature 338: 520-522.

Schneider, D.R, Fell, B.T., and Goldman, H., 1982, On the use of microwave radiation energy for brain tissue fixation, l Neurochem. 38: 749-752.

Seaman, RL., 1993, Modification of acoustic startle amplitude by microwave pulses, In: ElectriCity and Magnetism in Biology and Medicine, M. Blank:, ed., San Francisco Press, San Francisco, CA, pp 662-664.

Seaman, R.L., and Bebio, D.A., 1992, Modification of acoustic startle by microwave pulses in the rat: a preliminary report, Bioelectromagnetics 13: 323-328.

Seaman, RL., Beblo, D.A., and Raslear, T.G., 1994, Modification of acoustic and tactile startle by single microwave pulses, Physiol. Behav. 55: 587-595.

Seaman, R.L., and DeHaan, RL., 1993, Inter-beat intervals of cardiac-cell aggregates during exposure to 2.45 GHz CW, pulsed, and square-wave-modulated microwaves, Bioelectromagnetics 14: 41-55.

Seaman, R.L., and Wachtel, H., 1978, Slow and rapid responses to CWand pulsed microwave radiation by individual Aplysia pacemakers, l Microwave Power 13: 77-86.

Seaman RL., Belt, M.L., Doyle, lM., and Mathur, S.P., 1998, Ultra-wideband Electromagnetic pulses and morphine-induced changes in nociception and activity in mice, Physiol. Beh. 65: 263-270.

Seaman, R.L., Kiel, R.L., Parker, lE., Grubbs, TR, and Prol, R.K., 1996, Effects of ultrawide-band pulses on nitric oxide production in murine macrophages, In: Abstracts of the Bioelectromagnetics Society 18th Annual Meeting, Bioelectromagnetics Society, Frederick, MD, p 213.

Seaman, RL., Belt, M.L., Doyle, lM., and Mathur, S.P., Hyperactivity caused by a nitric oxide synthase inhibitor is countered by ultra-wideband pulses. Bioelectromagnetics 20: 431-439.

Sherry, C.J., Blick, D.W., Walters, T.J., Brown, G.C., and Murphy, MR, 1995, Lack of behavioral effects in non-human primates after exposure to ultrawideband electromagnetic radiation in the microwave frequency range, Radiat. Res. 143: 93-97.

Skidmore, W.D., and Baum, S.J., 1974, Biological effects in rodents exposed to 108 pulses of electromagnetic radiation, Health Phys. 26: 391-398.

Sparrow, M.G., Roggendorf, H., and Vogel, W.R., 1987, Effect of ethanol on heart rate and blood pressure in nonstressed and stressed rats, Life Sci. 40: 2551-2559.

Stevens, lC., 1983, Thermal sensation: infrared and microwaves, In: Microwaves and Thermoregulation, E.R. Adair, ed., Academic Press, New York, NY, pp 191-20l.

Stewart-DeHaan, P.I, Creighton, M.O., Jacobi, IR., Ross, W.M., Sanwal M., Guo, T.e., Guo, W.W., and Trevithick, lR, 1983, In vitro studies of microwave-induced cataract: separation offield and heating effects, Exp. Eye Res. 36: 75-90.

Stewart-DeHaan, P.l, Creighton, M.O., Larsen, L.E., Jacobi, lR., Sanwal, M., Baskerville, lC., and Trevithick, lR., 1985, In vitro studies of microwave-induced cataract: reciprocity between exposure duration and dose rate for pulsed microwaves, Exp. Eye Res. 40: 1-13.

262

Stodolnik-Baranska, W., 1974, The effect of microwaves on human lymphocyte cultures, In: Biological Effects and Health Hazards of Microwave Radiation, P. Czerski, Ch. Silverman, K Ostrowski, M.l Suess, M.L. Shore, and B. Waldeskog, eds., Polish Medical Publishers, Warsaw, Poland, pp 189-195.

Sutton, CR., and Carroll, F.B., Effects of microwave-induced hyperthermia on the bloodbrain barrier ofthe rat, Radio Sci. 14/6 Supplement: 329-334.

Taylor, E.M., and Ashleman, B.T., 1974, Analysis of central nervous system involvement in the microwave auditory effect, Brain Res. 74: 201-208.

Taylor, lD., 1995, Ultra-wideband radar overview, In: Introduction to Ultra-wideband Radar System, lC Taylor, ed., CRC Press, Boca Raton, FL, pp 1-10.

Thomas, lR., Finch, E.D., Fulk, D.W., and Burch, L.S., 1975, Effects oflow-Ievel microwave radiation on behavioral baselines, Ann. N.Y. Acad. Sci. 247: 424-432.

Thomas, lR., Schrot, l, and Banvard, R.A, 1982, Comparative effects of pulsed and continuous-wave 2.8-GHz microwaves on temporally defined behavior, Bioelectromagnetics 3: 227-235.

Tsong, T.Y., 1991, Electroporation of cell membranes, Biophys. l 60:297-306. Trevithick, lR., Creighton, M.O., Sanwal, M., Brown, D.O., and Bassen, H.I., 1987,

Risto-pathological studies of rabbit cornea exposed to millimeter waves, Proc. Ann. Conf. IEEE EMBS 9: 695-697.

Wachtel, H., Beblo, D., Vargas, C., Bassen, R., and Brown, D., 1988, Single microwave pulses can suppress startle reflexes in mice, ProC. Ann. Int. Com. IEEE EMBS 10: 911-912.

Wachtel, R., Beblo, D., Vargas, C., Bassen, H., and Brown, D., 1989, Comparison of the efficiency of pulsed versus CW microwave fields in evoking body movements, Proc. Ann. Int. Com. IEEEEMBS 11: 1136-1137.

Wachtel, R., Seaman, R., and Joines, W., 1975, Effects oflow-intensity microwaves on isolated neurons, Ann. N.Y. Acad. Sci. 245: 47-62.

Walters, TJ., Mason, P.A., Sherry, C.l, Steffen, C, and Merritt, lR., 1995, No detectable bioeffects following acute exposure to high peak power ultra-wide band electromagnetic radiation in rats, Aviat. Space Environ. Med. 66: 562-567.

Wangemann, R.T., and Cleary, S.F., 1976, The in vivo effects of2.45 GHz Microwave radiation on rabbit serum components and sleeping times, Rad. Environ. Biophys. 13: 89-103.

Williams, W.M., Hoss, W., Formaniak, M., and Michaelson, S.M., 1984a, Effect of2450 MHz microwave energy on the blood-brain barrier to hydrophilic molecules. A Effect on the permeability to sodium fluorescein, Brain Res. Rev. 7: 165-170.

Williams, W.M., del Cerro, M., and Michaelson, S.M., 1984b, Effect of 2450 MHz microwave energy on the blood-brain barrier to hydrophilic molecules. B. Effect on the permeability to HRP, Brain Res. Rev. 7: 171-181.

Williams, W.M., Platner, l, and Michaelson, S.M., 1984c, Effect of2450 MHz microwave energy on the blood-brain barrier to hydrophilic molecules. C Effect on the permeability to [14C]sucrose, Brain Res. Rev. 7: 183-190.

Williams, W.M., Lu, S.-T., del Cerro, M., and Michaelson, S.M., 1984d, Effect of2450 MHz energy on the blood-brain barrier to hydrophilic molecules. D. Brain temperature and blood-brain barrier permeability to hydrophilic tracers, Brain Res. Rev. 7: 191-212.

263

Williams, W.M., Lu, S.-T., del Cerro, M., Hoss, W., and Michaelson, S.M., 1984e, Effects of2450 MHz microwave energy on the blood-brain barrier: an overview and critique of past and present research, IEEE Trans. Microwave Theory Tech. 32: 808-818.

Wolke, S., Neibig, u., Elsner, R., Gollnick, F., and Meyer, R., 1996, Calcium homeostasis of isolated heart muscle cells exposed to pulsed high-frequency electromagnetic fields, Bioelectromagnetics 17: 144-153.

Zimmermann, u., 1982, Electric field-mediated fusion and related electrical phenomena, Biophys. Acta 694: 227-277.

264

A COMPREHENSIVE REVIEW OF THE RESEARCH ON BIOLOGICAL EFFECTS OF PULSED RADIOFREQUENCY RADIATION IN RUSSIA AND THE FORMER SOVIET UNION

Andrei G. Pakhomov1,2 and Michael R. Murphl

IMcKessonHBOC BioServices, U.S. Army Medical Research Detachment of the Walter Reed Army Institute of Research 2Directed Energy Bioeffects Division, Human Effectiveness Directorate, Air Force Research Laboratory Brooks Air Force Base, San Antonio, Texas, USA

INTRODUCTION

Pulsed radiofrequency (RF) radiation nowadays is among the most ubiquitous of environmental factors. Biological effects and health hazards of pulsed RF have been studied for decades, but there is still little consensus on whether the safety of pulsed RF should be given any special consideration compared to continuous wave (CW) emissions. While it is well known that pulsed fields can produce effects principally different from those of CW (e.g., an auditory effect, see Lin, 1990), potential implications of such effects for human safety and wellbeing continue to be debated.

A significant contribution to the field of bioelectromagnetics has been made by the research performed in the former Soviet Union (FSU). Unfortunately, most of this research was published in Russian; these publications are scarcely available in the West and have not ever been reviewed in English. Even some key findings, which may affect the conceptual understanding of interaction mechanisms and approaches to RF safety, seem to be not known in the West, and their replication in Western laboratories has never been attempted.

The goal of the present paper is to fill in this void and deliver a comprehensive review of the FSU research on biological effects and potential health hazards of pulsed RF radiations.

SOURCES OF INFORMATION

The principal source of information is a personal collection of scientific publications in Russian, which was accumulated as a part ofa 7-year effort of one of the authors (A.G.P.) to

Advances in Electromagnetic Fields in Living Systems, Vol. 3 Edited by le. Lin, Kluwer AcademiclPlenum Publishers, 2000 265

monitor and review the RussianlFSU research in the area of electromagnetic biology. Currently, this collection includes over 1200 titles and is one of the largest of its kind.

Over 1000 of the collected studies were abstracted in English, and the abstracts are included in the electronic EMF Database (produced and distributed by Information Ventures, Inc., PA, USA). Ibis Database (v. 3.2.4, 1999), which is the second major source of information for the current review, contains about 26000 unique citations of studies on bioeffects of electromagnetic fields (from DC to mm-waves). Out of them, there are about 3600 citations of studies performed in the FSU (and published in either Russian or English); we estimate that this bibliography includes at least 80% of related studies performed in the FSU and published in the open (unclassified) literature since 1975 (earlier papers are only scarcely included in the Database). The EMF Database proved to be extremely convenient as a fast reference and information search engine, while the scientific quality of studies usually had to be judged from the original publications.

STATISTICS AND GENERAL NOTES

For the purpose of this review, the term "pulsed RF" applies to any pulsed, intermittent, or amplitude-modulated electromagnetic emissions with carrier frequencies from units of megahertz up to 300 GHz. A detailed search of the EMF Database revealed a total of 206 citations of pulsed RF studies performed in the FSU. For convenience, they will be referred to below as "Russian" studies, disregarding the exact place of origin (e.g., Russia or Ukraine) or the language of publication. A similar search for "non-Russian" publications on the subject revealed 1342 citations. The percentage of pulsed RF publications out of the total number in the Russian and non-Russian bibliographies was remarkably the same (6%).

These 206 citations were individually sorted to remove all meeting abstracts, review papers, and papers on dosimetry, modeling, standards setting and enforcement; then we added appropriate Russian papers not cited in the Database. The final outcome was 114 original experimental Russian studies on pulsed RF bioeffects.

Of these studies, over 70% were performed in laboratory animals. Only a few studies «4%) employed pro- or eukaryotic cell cultures, whereas a lot of research was done in various isolated organ preparations (about 20%). A relatively large number of studies (>35%) explored effects of chronic irradiation.

It is generally accepted in Russian science that the nervous system is the most sensitive to electromagnetic radiation. Indeed, studies of pulsed RF effects on nervous system function and structure are by far the most numerous (>65%); electrophysiological and behavioral techniques are the most widely used. In a striking contrast with the Western research, not a single study has explored possible carcinogenic effects of pulsed RF; apparently, this possibility has never been a concern in Russian science.

Presumably nonthermal (or "RF-specific") bioeffects were found in 88 studies, which is more than 90% of all studies that searched for such effects. Among them, 34 studies compared bioeffects of various modulation regimens and/or those of CW versus pulsed RF. In most cases (>80%), it was found that pulsed RF effects indeed depend on modulation and/or are different from effects of CW irradiation at the equivalent time-average intensity. These data indicated that pulsing may be an important (or even the most important) factor that determines the biological effects of low-intensity RF emissions.

The most common question about Russian research in the bioelectromagnetic area is its scientific quality. Indeed, noticed flaws include unclear or invalid experiment protocols (e.g.,

266

lack of a sham-exposed control group); absence of good thermometry and dosimetry (in most animal studies, the specific absorption rate (SAR) was neither measured nor calculated); inadequate statistical analysis; failure to take into account specifics of biological experimentation with RF radiation (e.g., the use of metal recording electrodes in electrophysiological studies with RF could cause serious artifacts); and failure to report if all potential sources of the artifact were considered (e.g., the noise from a working RF transmitter can cause behavioral responses that may be misinterpreted as an RF bioeffect). While many studies were flawed in one way or another, or were poorly presented in publications, still many other Russian studies demonstrated a high-quality research.

For any study on RF bioeffects, whether flawless or not, the most critical issue is independent replication of findings. Until (or unless) proven wrong in replicative studies, the results of any reasonable-quality research deserve attention and careful consideration. That is why we attempted to address in this review all available Russian studies on pulsed RF bioeffects that meet at least the minimum scientific quality criteria. The other studies, which failed to meet these criteria (as far as it could be judged from the published material), were generally regarded as unacceptable. For example, all of the available Russian epidemiological studies with pulsed RF reported adverse health effects of exposure (such as "asthenic syndrome" or "radiofrequency disease"), but none of them showed reasonable evidence that these disorders were in fact caused by the RF exposure; besides, these studies reported too little or even no data about the exposure parameters. Because of the poor quality, none of the Russian epidemiological studies are included in the present review.

Among the "acceptable" studies, more credit was given to those that (1) reported interesting and potentially important findings, (2) presented data that were consistently reproduced over the years and were reported in more than a single publication, and (3) focused on research topics that were independently explored in different laboratories. The reviewed papers are arranged under the following categories: "In Vitro and In Situ Studies", "Animal Studies: Acute Exposure", "Animal Studies: Chronic Exposure", and "Bioeffects of Extremely High Power Pulses".

IN VITRO AND IN SITU STUDIES

Kim and co-authors (19S6) used fluorescent probes to reveal possible effect of pulsed microwaves (SOO MHz, 2 W/cm2, 50-~s pulses at 25-100 Hz) on the state of the cell membrane in erythrocyte ghosts. The temperature of exposed samples was kept stable at IS-19°C during 20-25 min of exposure. The experiments established that fluorescence of probes that bind to the lipid-water boundary of the membrane (2-toluidinonaphthalene-6-sulfonate, or 2,6-TNS, and l-anilinonaphthalene-S-sulfonate, or I,S-ANS) depended on the modulation frequency. Fluorescence of 2,6-TNS was not different from control at 25-35 Hz, but exceeded it by 12-16% at 55-65 Hz, and by 6-9% at SO-100 Hz. With I,S-ANS, the increase in fluorescence intensity at 55 Hz was 16.9 ± 5.1%. (Hereinafter, shown average values represent the mean ± standard error, unless stated otherwise.) Fluorescence increased to the maximum after 10-20 min of exposure and then remained stable; the increase was more pronounced at higher concentrations ofNaCI in the medium (up to 300 mmol). In contrast, the fluorescence parameters of piren, a hydrophobic probe, were not affected by irradiation. The authors supposed that the changes registered with fluorescent probes were induced by mechanical oscillations generated by microwave pulses.

267

Pashovskina and Akoev (1996) studied the effect of pulsed 2375 MHz radiation on the ATPase activity ofrat actomyosin in vitro. Exposures for 1 min at either 40 or 200 mW/cm2

(50- to 310-Hz modulation) heated the samples by less than 0.3 °C. ATPase activity was measured from accumulation of the inorganic phosphate (IP). Two parallel control samples, prepared together with the exposed one and from the same tissue homogenates, were used to establish the background IP contents at the time before and after the exposure. With 40 mW/cm2 radiation, 130- and 300-Hz modulation suppressed the ATPase activity to 50 ± 3% and 9.4 ± 2.7% of the control level, respectively (p<.05). In contrast, 270-Hz modulation increased the ATPase activity almost 3-fold (p<.05). Most of other tested frequencies produced weak or moderate « 50%) activity changes. The dependence of the effects upon the modulation frequency did not show any pattern or shape. Increasing the field intensity to 200 mW/cm2 entirely changed the effectiveness of the same modulation frequencies. Exposure at 130 and 300 Hz then caused ATPase activation (by 170.3 ± 3.3% and 61 ± 3.9%, respectively). The activation reached a maximum at 110-130 Hz modulation and decreased at higher and lower frequencies, forming a bell-shaped dependence. The authors concluded that ATPase activity of actomyosin showed a complex dependence upon both the field intensity and modulation, but were unable to explain this result.

Semin et ai. (1995) studied the effect of weak RF on the stability of DNA secondary structure in vitro. DNA was exposed in the presence of glycine and formaldehyde. Aminomethynol compounds, which form in this medium, react with DNA bases at singlestrand sites, which prevents recovery from damage to the DNA secondary structure. The damage accumulates during the incubation, and its amount can be estimated from the dynamics of thermal DNA denaturalization after RF or sham exposure. Samples were exposed for 30 min in an anechoic chamber at 18 °C by 10 different microwave frequencies simultaneously (4-to 8-GHz range, 25-ms pulses, 1- to 6-Hz repetition rate, 0.4 to 0.7 mW/cm2 peak power, no heating). Parallel control samples were sham exposed in a shielded area in the same chamber. The experiments established that irradiation at 3 or 4 Hz and 0.6 mW/cm2 peak power clearly increased the accumulated damage to the DNA secondary structure (P<0.00001). However, changing the pulse repetition rate to 1, 5, or 6 Hz, as well as changing the peak power to 0.4 or 0.7 mW/cm2, eliminated the effect entirely. Thus, the effect occurred only within narrow "windows" of the peak intensities and modulation frequencies.

One more example of a "double window" was described by Gapeev et ai. (1994). In this case, the effect required a certain combination of the carrier and modulation frequencies, but did not show a window dependence on the field intensity. The authors explored if the spontaneous locomotor activity of a unicellular organism Paramecium caudatum could be affected by irradiation with CW and modulated millimeter waves. Paramecia were placed in a thin layer of saline in a small round cuvette (7-mm diameter). Hyperpolarization of cells by a 5-fold reduction of potassium content in the saline (to 0.2 roM) made them constantly circle around in the cuvette for hours. A "motility index" (MI) was evaluated automatically by an optic probe as the amount of light reflected from cells appearing in the field of vision of a microscope during any 2-min interval. The Ml increased with increasing average locomotion velocity, and decreased when cells made more turns or when the number of active cells decreased. Radiation was applied for 12 min after a lO-min stabilization period. Control samples were monitored for the same time interval, but without irradiation.

268

A 42.2GHz

0.1 mW/cm 2

30 "#

~ 20 u w o

~ 10 ~

~ ...J

~ 0

, ! I

0.090 0.095 0.100

Hz

B 10mW/cm2

0.0956 Hz

I , ! ,

42.0 42.2 42.4 42.6

GHz

0.001

c 42.2 GHz

0.0956 Hz

0.1 10

mW/cm 2

Figure 1. Effects of pulsed millimeter-wave radiation on Paramecia motility index. A, the effect of modulation frequency (abscissa, Hz) when the carrier frequency and power are kept constant at 42.2 GHz and 0.1 mW/cm 2•

B, the effect of carrier frequency (abscissa, GHz) when the power and modulation are kept constant at 10 mW/cm2 and 0.0956 Hz. C, the effect of the incident power density (abscissa, mW/cm 2) at the resonance carrier and modulation frequencies (42.2 GHz and 0.0956 Hz). Dashed line shows the temperature increase, 0c. A single datapoint ("CW") illustrates the lack of motility changes for exposure without modulation at 20 mW/cm 2•

Each datapoint is the mean of 5-10 experiments; the error bars are confidence intervals at p<.05. See text for more detail. (Adopted from Gapeev et al., 1994).

In the first set of 30 experiments, cells were exposed at 42.2 GHz (CW) at power densities from units of microwatts to tens of milliwatts per centimeter square. These exposures never affected the MI, even at the highest intensities, which caused saline heating by 1-3 DC. In the next 30 experiments, the carrier frequency was modulated at 16, 8, 1, 0.5, 0.25, 0.1, or 0.05 Hz (0.1 to 20 mW/cm 2, 0.5 duty ratio). No effect was detected, except for the combination of 0.1 Hz modulation at 0.1 mW/cm 2, which considerably decreased the MI. More accurate study of the effect of various modulation rates revealed a clear resonant dependence with a peak at 0.0956 Hz (Figure 1, A). At this modulation rate, the threshold field intensity was near 0.02 mW/cm 2. The maximum effect of about 20010 MI reduction was achieved at 0.1 mW/cm 2, with no further increase at intensities of up to 50 mW/cm 2 (Figure 1, C). Further experiments demonstrated that the effect is induced only in a narrow window of carrier frequencies, with the peak at 42.25 GHz (Figure 1, B).

This effect could not be explained by microwave heating, which reached only 0.1-0.2 DC at 5 mW/cm 2 . Moreover, the effect could not be reproduced by heating with infrared radiation modulated at 0.0956 Hz. The authors suggested that the microwave-induced MI decrease could be caused by increasing the intracellular Ca2+ level and resulting cell depolarization. However, they were unable to explain reasons for the sharp resonant dependence of the effect upon both the modulation rate and radiation frequency.

A subsequent study by the same group of authors explored the effects of CW and modulated millimeter waves on the respiratory burst in murine peritoneal neutrophils (Gapeev et aI., 1997). Isolated neutrophils were exposed in the far field for 20 min in the

269

presence of a high concentration (7.5-10 J.LM) of calcium ionophore A23187. After the exposure, a respiratory burst was activated by adding phorbol-12-myristate-13-acetate (1 ~, and the production of reactive oxygen species (ROS) by neutrophils was measured from luminol-dependent chemoluminescence. Experiments with CW irradiation at 50 Il W Icm 2 at various fixed frequencies in the range 41.75-42.15 GHz revealed a 25% inhibition ofROS production at a "resonance" frequency of41.95 GHz; the half width of the resonance was about 100 MHz. At 41.95 GHz, a 50010 effect (half-maximum) was produced by intensities under 1 IlW/cm2. The effect reached maximum (about 24% inhibition) at 20 IlW/cm 2 and remained at this plateau level at higher intensities. Modulation of 41.95-GHz radiation at 1 Hz enhanced ROS production by about 10%, modulation rates of 0.1, 16, and 50 Hz inhibited ROS production by 16-20%, and the rates of 0.5, 2, 4 and 8 Hz had no effect. In the next experiments, the modulation rate was kept constant at 1 Hz and the radiation frequency was varied. This irradiation activated ROS production in the carrier frequency band of 41.95-42.05 GHz, and suppressed it in the band of 41.8-41.9 GHz. The authors concluded that the microwave effect on ROS production is nonthermal in nature, and is determined by specific combinations of the carrier frequency and modulation rate.

Tygranyan (1986) reported severe disturbances in isolated frog nerve and heart function if intense enough microwave pulses coincided with the "active" state of the excitable membrane (e.g., were applied during propagation ofthe compound action potential, CAP). In isolated frog sciatic nerve, these effects included decrease of the CAP amplitude (by 90-95%) and velocity (by 35-40%) after some 30 min of exposure. The same RF pulses delivered before, after, or asynchronously with CAP propagation produced only a subtle thermal effect. The author supposed that microwave pulse energy is converted into a mechanical pulse, which travels in a spiral between Schwann cell membranes of the nerve sheath. Due to a phase synchronism effect, the traveling pulse triggers another mechanical pulse perpendicular to the nerve surface, and the latter may be strong enough to damage the "active" nerve membrane. The author's calculations based on dielectric and mechanical properties of nerve seemed to support this hypothesis.

The findings of Tygranyan motivated Pakhomov et al. (1991) to search for similar effects in giant nerve fibers of the isolated ventral nerve cord of the earthworm Lumbricus terrestris. A length of about 2.5 mm of the cord was exposed in a small drop of Ringer's solution to 6.45-GHz CW or pulsed radiation, or was sham exposed. Exposure duration was from 10 to 50 min at the peak SAR from 30 to 230 mW/g. Action potentials, which were evoked by repetitive electrical stimulation and propagated along the nerve, were recorded by two pairs of electrodes, positioned immediately before and after the exposure zone. Therefore, comparison of these records provided an accurate measure indicating whether the nerve conduction in the exposure zone was affected. Microwave pulses were synchronized with the action potential crossing the exposure zone, or were delivered independently from the nerve stimulation (0.2- to 32-ms pulse width, 6- to 2000-Hz repetition rate). This study failed to produce any specific microwave effect, regardless of RF pulse parameters or their phasing with the nerve firing. However, giant nerve fibers of the earthworm ventral cord do not have a myelin sheath that was theoretically supposed to be essential for Tygranyan's effect described above. The design of the exposure cell and the carrier frequency were also different (6.45 GHz versus 800-915 MHz).

For the next study (pakhomov et al., 1993), a microwave transmitter and irradiator were borrowed from Tygranyan's laboratory, to replicate his exposure conditions with maximum care. The experiments were performed in isolated frog sciatic nerve, which was continually

270

stimulated at a rate of 50 twin pulseslsec for 90 min. This high-rate stimulation caused gradual decline of the nerve function, i.e., increased the CAP latency and decreased its amplitude and tracing integraL Exposures or sham exposures began 1 min after the onset of the high-rate stimulation and continued till the end of the experiment. Microwave pulses (915 MHz, 5 to 70 Wig peak SAR, 0.5- or 3-ms width) were phased with CAP in various ways, or were delivered asynchronously at a rate of 50 Hz. At the highest SAR, microwave heating could reach 2.7 °C. Regardless of phasing, SAR, or pulse width, none of the exposure regimens caused severe CAP suppression, therefore the results of Tygranyan (1986) failed to be independently confirmed.

At the same time, these experiments revealed a different microwave effect: The exposed nerves displayed a faster decrease of the CAP amplitude and tracing integral during the course of the high-rate stimulation. This effect was regarded as nontherma1 or microwavespecific, since conventional heating of the superfusing solution by 3 °C during sham irradiation caused the opposite changes. Within studied limits; the magnitude of the effect showed no significant correlation with exposure parameters. In general, this microwaveinduced facilitation of the nerve vitality loss was similar to the effect reported earlier by McRee and Wachtel (1980, 1982), but substantially weaker.

Two subsequent studies explored the dependence of this effect upon modulation (Pakhomov et aI., 1992; Pakhomov, 1993). The course of the high-rate stimulation and exposure was reduced to 20 min. Microwave pulses of 1- to 1000-f.ls duration (915 MHz, 43-48 or 20-33 Wig peak SAR) were delivered at duty cycles from 1120 to 114000, and were not synchronized with nerve firing. These studies identified effective and ineffective pulsing regimens. For instance, at 1/40 duty cycle and 43-48 Wig, 1- and 100-f.ls pulses caused a statistically significant effect, while 10- or 1000-f.ls pulses did not, despite the fact that the gross microwave heating by all these exposures was essentially the same.

Presumably a nonthermal effect of pulsed RF was observed in in situ frog heart by Grechko (1994). This wide-scale study used 1680 male frogs to compare the effects of five different types of electromagnetic fields on cardiotoxic effectiveness of a drug strophantine K. Injection of this drug caused a characteristic heart arrest in systole in about 60 min. Irradiation by 3085 MHz, I-f.ls pulses at 400 Hz, 0.3-0.5 mW/cm 2 for 30 min significantly increased the latency of the heart arrest (by 17-44%). In some cases, the heart was even able to restore the activity, which never happened under sham exposure conditions. In contrast, the same irradiation for 10 min at 2-3 m W/cm 2 decreased the heart arrest latency by 10-25%. Pakhomov et aI. (1995b) attempted to replicate findings of Tinney and co-authors (1976) that microwaves can cause beating rate changes in isolated heart by stimulation of neurotransmitter release from autonomous nervous system intracardiac fiber terminals. CW exposure at 2-10 mWlg (960 MHz) was reported to cause bradycardia in isolated turtle hearts. This effect could be blocked by a parasympathetic blocker atropine, or enhanced by a sympathetic blocker propranolol. The replication experiments were performed in isolated frog heart slices, using 915 MHz radiation; exposures continued for up to 40 min at SAR levels from 0.1 to 52 mWlg (10- to 100-f.ls pulses at 16- to 2000-Hz repetition rates). Neither irradiation alone nor in combination with atropine or propranolol caused bradycardia in the frog heart slices. The only effect detected was a reversible beating rate increase when microwave heating exceeded 0.1 °C (when SAR exceeded 10 m Wig). One needs to note, however, that the studied drugs alone (i.e., without irradiation) caused marked beating rate changes in isolated turtle hearts (Tinney et aI., 1976), but not in the frog heart slices, even at much higher concentrations. This finding may reflect a principal difference in physiological

271

organization of these two heart preparations, which could be a reason for their different sensitivity to microwave radiation.

It was also reported that a caffeine antagonist, tetracaine, eliminated a nonthermal microwave effect in isolated snail neurons (Arber and Lin, 1984). Therefore, Pakhomov et al. (199Sb) inferred that caffeine itself could enhance the microwave effect and tested this hypothesis in isolated frog heart slices. In contrast to atropine and propranolol, superfusion of slices with 1 mM of caffeine strongly increased the average heart power, which was calculated as a product of the beating rate and beats' amplitude. Exposure at 8-10 mW/g average SAR (1.5-ms pulses, 2.S-ms interpulse interval, 8 pulses/burst, 16 bursts/s, 91S MHz for 33 min) augmented the caffeine effect by about IS% (p<.02). Microwave heating was under 0.1 °C and could not account for this exposure effect. Moreover, the same exposure without caffeine superfusion caused no beating rate or amplitude changes.

Afrikanova and Grigoriev (1996) reported that pulsed, but not CW exposure increased the probability of heart rate changes and heart arrest in isolated and in situ frog heart. The effect was observed at intensities from 16 to 340 IlW/cm2 at 9.3 GHz and modulation frequencies under 100 Hz (30% modulation depth, SO% duty cycle). The maximum effect was produced by a S-min exposure with modulation frequency changing by I-Hz steps from 6 to 10 Hz (1 min at each step). A longer exposure with this modulation (10 or 19 min), as well as S-min exposure using various other modulation patterns, produced weaker or different effects.

Pulsed RF effects on individual neurons in isolated brain slices were explored in studies of Zakharova and co-authors (1993, I99S, 1996). Spontaneous firing of cortical neurons was recorded by an extracellular microelectrode before, during, and after a brief (4-S min) 900-MHz exposure at 1.4 mW/g. The reactions observed with 7-, 16-,30-, and 60-Hz modulation (liS duty ratio) were a decrease of the firing rate and desynchronization of firing of individual neurons. The probability of the firing rate inhibition and its magnitude were the highest with 7- and I6-Hz modulation and the lowest with the 60-Hz modulation. For instance, 16-Hz pulses inhibited the activity in IS out of 17 tested neurons; the firing rate could fall by 24% on the 1st minute of irradiation, and by as much as 6S% on the 4th min. The rate did not recover but sometimes even decreased further after exposure. A far more intense conventional heating (by 0.33 °C/min, as opposed to 0.02 °C/min under exposure) also inhibited the activity, but just in 7 out of 13 neurons, and the firing rate did recover when the heating was over. The greater effectiveness of microwave treatment, its dependence on modulation, as well as lack of the recovery after exposure, suggested a specific microwave effect mechanism.

Philippova et al. (1988) reported an interesting effect of 900-MHz exposure on 3H_ camphor binding in isolated membrane fraction of rat olfactory epithelium. Under constanttemperature conditions, irradiation markedly decreased specific 3H-camphor binding (down to 60-40% of the control), but did not affect its nonspecific binding. The effect depended on SAR and duration of exposure, but not on modulation in the studied range from 1 to 100 Hz. At 1 mW/g, the effect gradually increased and reached saturation after some 10-20 min of exposure. For a IS-min exposure at various SARs, the effect gradually increased with SAR and reached saturation at about 4 mW/g. The authors supposed that the observed decrease of ligand binding resulted from microwave-induced release of specific camphor-binding protein from cell membranes into solution.

The microwave effect on olfactory epithelium and neighboring respiratory epithelium was further investigated in a histological study by Popov et al. (1996). Tissue samples were

272

obtained from Wistar rats; one sample from each animal served as a control, and the other one was exposed for 15 min at 15 Wlkg to 0.9 GHz microwaves (16 Hz pulsed modulation, 50% duty ratio). Subsequent electron microscopy revealed increased vacuolization of olfactory neurons in exposed samples, which indicated dying of these cells. The most severe degenerative changes (swelling, appearance of vesicles and multilamellar structures) developed in non-myelinated axons of olfactory neurons. Concurrently, support cells displayed pronounced cytoplasmic vacuolization in apical parts of the cell body, which indicated that microwaves enhanced mucus secretion. In the respiratory epithelium, the most remarkable effect was fusion of cilia into so-called "giant cilia," which contained numerous basal bodies. These bodies formed axonemes with a typical arrangement of microtubules. Giant cilia included 5-10 or more axonemes with basal bodies and numerous vesicles (remains of fused membranes). As a rule, the cytoplasm of respiratory cells was strongly vacuolated. Overall, the authors regarded these and other observed changes as a "direct and pronounced degenerative effect of microwaves on neurons and other epithelial cells."

To complete this section on in vitro bioeffects, we will mention the studies that attempted to demonstrate specific bioeffects of pulsed RF, but could find either no effects, or only the effects caused by heating. Alexeev et ai. (1987) could not find any but thermal effects on Ca2+ transmembrane current in mollusk neurons (900 MHz, <10 min at 0.2-20 mW/g, 0.5- to 1,000-Hz modulation). Khitrov and Kakushkina (1990) observed solely thermal effects of CW or pulsed exposure on K+ transport and oxygen consumption in isolated rat liver tissue (2450 MHz, 0.1-5 Wig, 13-, 17-,21-, or 25-Hz pulses at 111000 duty ratio). Khramov et al. (1991) could detect only thermal effects of microwaves on the electric activity of a crayfish stretch receptor (37- to 78-GHz, 10 to 250 mW/cm2, from 20 sec to several hours, 0.01-1000 Hz modulation). In a study by Bolshakov and Alexeev (1992), pulsed but not CW radiation produced bursting responses of snail neurons (900 MHz, 0.5 to 15 mW/g, 0.5 to 110 Hz); however, the authors regarded this effect as possibly being an artifact from mechanical vibrations of the exposure chamber. In experiments with isolated frog heart slices, Pakhomov et al. (1995a) tested over 400 modulation regimens and concluded that changes in the beating rate and amplitude are entirely determined by the timeaverage SAR and microwave heating (885 or 915 MHz, 1-l.I.s to 10-ms pulses, 117 to 11100,000 duty cycle, 100 to 3000 mW/g peak SAR for 2 min).

ANIMAL STUDIES: ACUTE EXPOSURE

The possible impact of low-intensity pulsed microwaves on natural ecosystems was studied in a laboratory model using a tick Hyalomma asiaticum (Korotkov et ai., 1996). Exposure regimens tested in this study differed in the microwave frequency, peak power, and modulation (see Table 1). Ticks were kept at a room temperature of 21-23 °C in humidified chambers. Exposures were performed at different stages of tick development (eggs, larvae, nymphs, imago); the number of specimens in each exposed group varied from 20 (experiments with larvae) to as many as 3000 (experiments with eggs). Irradiation of eggs 5-6 days after they were laid did not affect the percent of appearing larvae (95-100% in any group). At the same time, exposures significantly delayed larvae hatching. In the control

group (SI), 50% of larvae appeared in 11.9±1.0 days, while in the exposed groups this

interval increased to 15.3±0.1 days (R2), 25.3±0.1 days (R3), and 32.2±0.7 days (R4); R5 and R6 were not tested in these experiments. Activation of the larvae of S 1- and R2-exposed

273

groups occurred on the day of hatching, while R3 and R4 exposures delayed it by 17.0±O.4

days and 23.9±O.5 days, respectively. Exposures R4 and R5 decreased the life span of unfed larvae and nymphs by 20-30%, but this effect was not statistically significant.

Effects ofR3-R6 exposure regimens were studied in replete larvae in winter and fall tick generations. The larvae were exposed on the 10th day after feeding. The first nymphs appeared in 16-18 days after the feeding in all the groups, but their survival, molting, and life span were adversely affected by exposures. On the whole, irradiation by pulse bursts at 3 GHz (R3 and R4) was the most effective, and the effects were more profound at the higher peak power (R4). On the contrary, the wide-band exposures (R2 and R5) were the least

effective, even at 750 IlW/cm2 peak SAR. Active and fed ticks were less sensitive to microwaves than hungry specimens or those in the diapause period. The authors concluded that low intensity pulsed microwaves were a negative environmental factor for the tick H. asiaticum.

Table. 1. Microwave exposure regimens* employed in the study by Korotkov et al., 1996

Regimen Frequency, Peak power, Modulation GHz JlW/cm2 Pulse width,.i Pulse repetition iBurst duratiOnJ Burst repetition

ms rate, Hz ms rate, Hz

SI Sham - - - - -R2 1-4 150 20 7 - -R3 3 75 •• 1000 20 7

R4 3 150 ** 1000 20 7

R5 1-4 750 20 2 - -R6 1 750 20 2 - -

* Exposure duration in all cases was 15 min (Grigoriev, 1999). ** Pulse width for regimens R3 and R4 is not specified in the paper.

All other in vivo studies of acute bioeffects of pulsed RF were performed in warmblooded laboratory animals. The vast majority of these studies focused on changes in the nervous system and behavior; the other ones dealt with general physiological changes (e.g., hormone production, blood composition, heart rate), and just one study explored possible genetic damage from RF irradiation (Belokhvostov et al., 1995).

In this work, male albino rats (150-200 g) were exposed in an anechoic chamber to 1-

Ilsec, 400-Hz microwave pulses (3.085 GHz) at 5, 10, 20, or 46 mW/cm2. The exposure lasted for 24 min or 2 hours, and the authors claimed that it caused no rise in the rectal temperature. The animals were decapitated 5 hours after the exposure, and blood taken from 6-7 rats was pooled together and considered as one sample. Blood cells were removed, and the samples were analyzed by DNA electrophoresis in polyacrylamide gel and by polymerase chain reaction (PCR). Low-molecular-weight DNA fraction (175-185 nucleotide pairs) was revealed in blood plasma in 38% of the samples from sham-exposed animals (6 samples out of 16). In exposed groups, regardless of exposure duration and intensity, this fraction was found in 80 to 100% of samples (e.g., 7 out of8 samples after 24 min at 5 mW/cm2, see also Figure 2, A). The DNA fraction was analyzed further by PCR to reveal complete and incomplete copies of so-called LINE elements (which are a known marker of genetic

274

transposition in various pathologic conditions). Complete copies of the LINE element were not found in any control samples, but were present in lout of 7, 5 out of 7, and lout 7 samples after a 2-hour exposure at 5, 10, and 20 mW/cm2, respectively (i.e., 10 mW/cm2 was the most effective). The amount of DNA was, on the average, 2.24 times higher after 10 mW/cm2 exposure than after 20 mW/cm2 (n=3). Since the transposition of full bopies of the LINE element is known to cause DNA rearrangements within the genome, their release into blood plasma may indicate a possible adverse effect of microwaves on genome stability.

In contrast to complete LINE elements, partial copies of the LINE sequence (with 3'-end fragment and the middle portion of the sequence) were present in all samples taken from the exposed animals. For some sequence clones, the amount of DNA was proportional to the intensity of irradiation. For example, for the clone 43 (the sequence is given in the original article), the amount of DNA increased as 1.0/1.85/1.92/2.98 in the series shaml5110/20 mW/cm2 (Fig. 2, B). The increased presence of the partial LINE copies does not indicate any pathological processes, but may be a useful index for biological dosimetry of microwave exposure.

A 8

- 434

- 267

- 184

- 123

2 1 2 1 M 2 3 3 4 1

Figure 2. A, Presence of a low-weight (175-185 nucleotide pairs) DNA fraction in blood plasma of rats after pulsedRF exposure. 1: sham exposure, 2: 20 mW/cm2 for 2 hours, M: molecular weight marker. Values to the right show the number of nucleotide pairs in marker DNA. B, Revealing of the 3'-end fragment of the LINE DNA element (clone 43) by 25 cycles of peR amplification. 1: sham exposure, 2: 5 mW/cm2, 3: 10 mW/cm2,

and 4: 20 mW/cm2 • See text for more detail. (Adopted from Belokhvostov et aI., 1995).

Microwave exposure of the head of rabbits for 1 min at 0.2 mW/cm2 (6 GHz, 50-Hz pulses at 112 duty cycle) changed the spontaneous firing rate of individual neurons in the sensomotor cortex (Lukianova et al., 1995). The firing rate was recorded extracellularly with glass microelectrodes (5- to 20- Mohm resistance). The microelectrode holder and

275

micromanipulator were affixed to the animal's skull. These devices were custom made of plastic, to prevent any artifacts arising from metal objects introduced into the microwave field. For the same reason, plastic tubes filled with saline were used instead of metal wires to connect the microelectrode holder with an amplifier. The activity of 53 neurons was analyzed in 67 3-min experiments; each experiment was a continual activity recording for 1 min before, 1 min during, and 1 min after irradiation. The data were compared to those recorded from 54 neurons in control experiments with sham irradiation. During the exposure or sham exposure, the firing rate of an individual neuron could either increase or decrease, or showed no statistically significant changes. In the control experiments, the firing rate decreased in 8.2% of neurons and increased in 13.07%. In the experiments with irradiation, these numbers changed to 49.5% and 6.4%, respectively (p<.05). Thus, the most characteristic effect of exposure was suppression of the neuron firing. This suppression developed with an average latency of 12 sec after the onset of exposure. Other parameters studied, such as spike amplitude or firing pattern, were not affected by microwaves.

Table 2. Changes in "spontaneous" firing rate of neurons in rabbit cerebral cortex caused by a I-min sham or microwave exposure. (Adopted from Moiseeva, 1996, and Lukianova and Moiseeva, 1998)

Area of the Exposure Number of Type of response, % of the total number of neurons cortex regimen neurons Significant cbanges in the firing mte I Lack of

Rate increase I Rate decrease I changes

Sensomotor Sham 72 2.7 11.1 86.2

Pulsesl 105 27.6* 32.4* 40.0*

Bursts2 63 24.2* 31.0* 44.8*

Occipito- Sham 73 5.5 11.0 83.5 parietal Pulsesl 84 25.0* 29.8* 45.2*

Bursts2 63 23.2* 30.5* 46.3*

1 1.5 GHz, 0.3 mW/cm2 peak power, 0.4-ms pulses, 1000 pulses/sec. 2 1.5 GHz, 0.3 mW/cm2 peak power, 0.4-ms pulses, 1000 pulses/sec within 16-ms bursts, 0.12 burstslsec. * The percentage of neurons is significantly different from sbam control (p<.0 I, ·l test)

Similar experimental techniques and protocol were employed by Moiseeva (1996) and by Lukianova and Moiseeva (1998) to study the effect of 1.5 GHz microwaves. These two papers described effects of two different pUlsing regimens in an experiment with 22 rabbits. For both the regimens, the pulse width was 0.4 ms and the peak power was 0.3 mW/cm2•

These pulses were either continually delivered at a 1000-Hz repetition rate, or were arranged in 16-ms bursts (0.12 bursts/sec, 1000 pulses/sec within bursts). The data were collected from a total of 240 neurons in the sensomotor cortex and 220 neurons in the occipito-parietal cortex. Lack of statistically significant firing rate changes was the predominant response to sham exposure, while microwave irradiation markedly increased the percentage of neurons that either decreased or increased their firing rate (Table 2). Overall, about 60% of neurons reacted to microwaves. Both tested irradiation regimens caused a practically identical response. As a rule, exposure suppressed the activity in neurons with initially high firing rate, and facilitated the activity in neurons with initially low firing rate. The magnitude of changes

276

also was higher under microwave exposure. For exposure with repetitive microwave pulses, for example, the pre-exposure firing rate in the sensomotor cortex was 6.1 ± 0.89 pulses/sec; it increased to 11.2 ± 0.64 pulses/sec under irradiation (p<.05), and returned to 9.2 ± 0.58 pulses/sec after the irradiation. The respective numbers for sham exposure were 5.2 ± l.32, 6.94 ± 1.5, and 7.8 ± l.09 pulses/sec (the changes are not statistically significant). The average latency of the response was 10 ± 3 sec for firing activation, and 9 ± 2 sec for firing suppression.

Pulsed RF effects on brain receptors have been continually studied for more than 10 years (Akoev et al., 1985, Kuznetsov et al., 1991, Kolomytkin et al., 1994, Iurinskaia et al., 1996). A brief exposure of rats to 800, 880, or 915 MHz microwaves increased radio labelled agonist binding to glutamate receptors, decreased binding to gamma-amino butyric acid (GABA) receptors, and decreased acetylcholinesterase activity in brain. These effects required 16-Hz modulation, at either 85% or 32% duty ratio; CW irradiation or other modulation frequencies (3 to 30 Hz) caused no statistically significant changes. Interestingly, brief (1- and 5-min) exposures often produced greater effects than prolonged exposures (15-60 min). For a 5-min exposure with 16-Hz modulation, the threshold field intensity was between 10 and 50 /!W/cm2. The magnitude of the effects gradually increased or remained almost unchanged for the field intensities from 0.1 to 1 mW/cm2.

These findings can be illustrated in more detail by the paper ofIurinskaia et al. (1996). Four separate series of experiments were performed in male Wistar rats (150-200-g body weight). Animals were kept in normal vivarium conditions, and were made accustomed to experimental manipulations, environment, and handling for 7-10 days prior to experiments. All experiments were performed at the same time of day. The animals were exposed in groups of 3 in a plastic cage, from an open end of a waveguide. Control animals were treated in exactly the same manner and were subjected to a sham exposure. Rats were decapitated immediately after irradiation, and the brain was extracted and processed to evaluate the specific binding of radio labelled agonists CH-glutamate and 3H-muscimol) to glutamate and GABA receptors. In the first set of experiments, the authors studied the dependence of binding on the incident power density (from 0.01 to 1 mW/cm2), when other exposure parameters were kept constant (915 MHz, 16 Hz modulation for 5 min). Microwaves increased binding of 3H-glutamate and decreased binding of 3H-muscimol. At 10 /!W/cm2, these indices changed to 120 ± 12% and 88 ± 12%, respectively (sham control was taken as 100%). At higher power densities, both effects gradually became more substantial (over 200% and 70-80%, respectively) and statistically significant (Figure 3, A).

The next series of experiments was focused on determining the role of modulation in the induction of these effects. Animals were exposed for 5 min at 1 mW/cm2 (800 MHz for experiments with 3H-muscimol and 915 MHz for experiments with 3H-glutamate), modulated at 0, 2.5, 3, 5, 7, 16, or 30 Hz (32% duty ratio). Neither CW nor modulated regimens produced a statistically significant change in 3H-muscimol binding, except for 16-Hz modulation which decreased binding to 70 ± 5%. Maximum increase in the 3H-glutamate binding (to 200-220%) also occurred at 16-Hz modulation (Figure 3, B).

In the third set of experiments, the exposure duration was varied from 1 to 60 min, keeping the other parameters constant (800 or 915 MHz, 16 Hz modulation, 1 mW/cm2). Maximum decrease in the 3H-muscimol binding (to 45-50%) occurred after a I-min exposure, and the effect gradually weakened for longer exposures. For 3H-glutamate binding, the maximum effect of 200-220% was observed after a 5-min exposure; after I-min

277

exposure, it was 130 ± 6%. The ability to produce the effects with very low intensities of radiation, as well as dependencies on modulation and exposure duration indicated a nonthermal mechanism of the effects.

200

%

100

01

A 16 Hz, 5 min

t== ............................. . Mu + + +

10 100 )J.W/cm2

~ooo

B 1 mW/crn 2, 5 min

Figure 3. Changes in the binding activity of glutamate and GABA receptors in the rat brain after exposure to modulated 915-MHz microwaves. A, the effect of a 5-min exposure with 16-Hz modulation on binding of

radiolabelled muscimol (Mu) and glutamate (Glu) at different incident power densities (abscissa, IlW/cm\ B, the effect of a 5-min, 1 mW/cm2 exposure on glutamate binding at different modulation frequencies (Hz). The

data shown are the mean ± s.e., in % to sham-exposed controls. See text for more detail. (Adopted from Iurinskaia et aI., 1996).

In the forth set of experiments, the authors studied microwave effects on the concentration dependences of the change in binding characteristics. Irradiation only slightly affected the dissociation constant of muscimol, but decreased the number of binding sites from 17.4 ± 0.8 to 10 ± 2 pmol per mg of protein. In contrast, irradiation did not change the number of binding sites for glutamate, but decreased its dissociation constant from 277 ± 15 nM (control) to 103 ± 10 nM.

Summarizing their experimental results, the authors noted that the dynamics of the changes they observed in the state of brain receptor systems is similar to the effects produced by an electric shock, pain, or immobilization stress. They hypothesized that electromagnetic radiation works as a stressing factor, with adaptation developing under a prolonged exposure.

A series of studies performed under the leadership of K. V. Sudakov explored neurophysiological and behavioral effects of the lower RF frequencies (30 to 40 MHz, 30 to 300 Vim, 2- to 50-Hz sinusoidal modulation). The experiments were performed in rats and rabbits, and exposures lasted from 2-5 min to 45-90 min. These exposures inhibited the autonomous nervous system response to stimulation of the hypothalamus (Kashtanov and Sudakov, 1981); alleviated stress response to immobilization and electrical stimulation of the skin (Gorbunova et aI., 1981); induced spreading epileptiform activity in the brain, followed by EEG suppression; induced the appearance of slow waves and even catalepsy in some

278

animals (Sudakov and Antimonii, 1977); suppressed defensive behavior and motor activity, and inhibited extinction of the conditioned response (Sudakov, 1976, 1998). Depending on the modulation frequency, the exposure had a biphasic effect on the self-stimulatory behavior in rats (activation followed by suppression at 2-Hz modulation, or no effect followed by suppression at 7-Hz modulation), or even suppressed the self-stimulation practically from the onset of exposure at 50-Hz modulation (Antimonii et al., 1976).

However, these interesting findings should be taken with caution. One reason is that, in many cases, the authors used regular metal electrodes for brain stimulation and recording of the electric activity. The use of such electrodes during the electromagnetic field exposure could cause field distortion and various artifacts (e.g., demodulation of the field and direct stimulation of biological structures by electric current at the modulation frequency). Without independent replication in artifact-free conditions, it is not possible to judge if these findings were in fact the effect of the field, or were just a result of the flawed experiment protocol.

Konovalov and Andreeva (1989) studied microwave effects on apomorphine-induced motor stereotypy in rabbits and rats. In rabbits, a 30-min exposure of the head (880 MHz, 8 mW/cm2, 16- or 30-Hz pulses or CW) decreased the number of "simultaneous hits" by hind limbs by 15-25%, but this difference was not statistically significant. The authors noted, however, that in 3 animals (out of 34) 16-Hz pulses decreased the number of "simultaneous hits" 3-4 times, while the other exposure regimens had only a subtle effect. In rats, a 1 mg/kg dose of apomorphine caused them to circle around the cage in the left or right direction. A single 60-min whole-body exposure with 16-Hz pulses suppressed this behavior by 21% (p<.05 compared with sham exposed controls); a series of 5 exposures for 60 min/day suppressed it by 40% (p<.OI).

Grigoriev and Stepanov (1998) studied delayed effects of pulsed RF exposure on imprinting behavior in chicks. Eggs were exposed on the 16th day of incubation, for 5 min at 40 IJ.W/cm2 (10 GHz, 1-,2-,3-, 7-, 9-, or 10-Hz modulation). Two days after hatching, chicks were given a choice of two flashing light stimuli. The flashing rate of one light was the same as the modulation rate in the previous RF exposure of the egg; the flashing rate of the other light was offset by 8 Hz. It was found that chicks exposed in eggs to 9- or 10-Hz RF pulses showed preferences to the light flashes of the same frequency. Sham-exposed chicks and those exposed at other RF pulsing rates did not show a preference to either light stimulus.

A single 50-min microwave exposure at very low average intensity (15 IJ.W/cm2, 6 GHz, 2-Hz pulses) produced no statistically significant changes in the physiological condition of adult rabbits (Rynskov et al., 1995). The condition of microwave- and sham-exposed animals was evaluated immediately after the exposure and also on the next day. The tests employed in this study were electrocardiogram, electroencephalogram, compound myogram of musculus gastrocnemius, pneumogram, and blood contents of cortisol, testosterone, insulin, and thyroxin.

Lukianova and co-authors (1995) compared formation of conditioned avoidance reflexes to light, sound, pulsed microwaves (6 GHz, 50-Hz pulses, 200 IJ.W/cm2 average and 400 IJ.W/cm2 peak power), and 1000-Oe constant magnetic field. Rabbits were taught to respond to only one type of the conditioning stimuli (3 animals per group), and 4 other animals were subjected to "sham stimuli". Exposures and sham exposures were performed 6 times every other day, up to the total of 400 treatments for each animal. In 20 sec after the onset of the conditioning stimulation, the rear left extremity was stimulated with electric current, causing a flexor reaction. Withdrawal of the leg within this 20-sec interval prevented the electric stimulation and was regarded as a conditioned reflex response. Consolidation of the

279

conditioned reflex was observed in all but one exposed animal and in none of the shamtreated controls. Consolidation took noticeably more time in the case of magnetic field exposures (243-245 combinations of stimuli) than with exposures to microwaves, sound, or light (139-154, 98-134, and 114-184 combinations, respectively). The maximum reflex percentage per 50 combinations established for the above groups was 40-60%, 30-60%, 60-78%, and 70-80%, respectively. The average latency of the reflex to the magnetic field (8.5 ± 0.12 sec) and to microwaves (7.76 ± 0.35 sec) significantly exceeded the corresponding numbers for sound (4.9 ± 0.09 sec) and light (4.47 ± 1.17 sec).

Navakatikian (1992) has described a behavioral technique sensitive enough to detect the effect of acute pulsed RF exposure at 10 JlW/cm2 (or 2.7 JlW/g). The author employed a well-known shuttle-box method of studying the conditioned reflex of active avoidance, but scrupulously refined the shuttle box design and optimized the schedules of animal training and testing. Particular attention was given to selection of uniform experimental and control groups and to statistical analysis of the animals' performance data. An experiment with microwave exposure illustrated all the detail of how to use this behavioral method and how to process the raw data.

Adult male rats (280- to 360-g body weight) were trained to respond to a sound stimulus by relocating from one compartment of the shuttle box into another, in order to avoid punishment by electrical stimulation. The optimum training schedule consisted of three training series, each with 75 combinations of stimuli. One additional, so-called "testing" series (identical to the previous training series) was performed immediately after a 30-min sham or microwave exposure (3000-MHz, 10 JlW/cm2, 400-Hz pulse repetition rate, 2-Jlsec pulse width, 4 pulses per burst, one burst every 3.75 sec). The difference in each animal's performance in the testing series and in the last training series was taken as an individual change (IC) index. The IC values were averaged over groups of 11 exposed and 11 control rats. For the latency of the reflex response, the average IC was -0.11 ± 0.06 sec for the shamtreated group, and +0.17 ± 0.09 sec in the exposed group (p<.05). The author noted that, for an acute exposure, the minimum microwave intensity that was reported in literature to change conditioned reflexes is 1000 JlW/cm2. The method described in this paper has therefore been proven to be sensitive enough to establish an effect of acute microwave exposure at an intensity two orders of magnitude less than previously reported.

However, the significance of this demonstration should not be overestimated. The reason is, that with the employed pulsing regimen, the incident energy density per pulse reached 9 JJ.1/cm2, which clearly exceeds the established threshold of 1.5-3 JJ.1/cm2 for auditory perception ofRF pulses by rats (Chou et al., 1985). Therefore, it is not surprising that a sensitive behavioral technique could reveal some effect of a 30-min RF auditory stimulation. On the other hand, the author has reported elsewhere (Navakatikian, 1993) that the described method reliably detects the effect of 2450 MHz CW exposure at 0.1 mW/cm2

(27 JlW/g); this finding suggests that the microwave effect does not necessarily involve just auditory or thermal stimulation.

ANIMAL STUDIES: CHRONIC EXPOSURE

A substantial number of studies have explored behavioral and physiological effects of exposure conditions that imitated actual occupational or residential exposures from various military and civilian radar systems. The employed modulation schemes could be rather

280

complex, such as regular bursts of pulses from rotating antennas or field modulation by Morse code.

Potential health effects of a coastal radar were explored by Tomashevskaia and Solenyi (1986). Field measurements established that popUlation within O.l-l-km distance from the coastal radar can be exposed to intermittent bursts of 400- or 800-Hz microwave pulses (3- or 10-cm wavelength) at levels from 10 to 80 JlW/cm2; the intermittence was produced by antenna rotation at 16 rpm. White mongrel rats (130 -140 g body weight) were exposed to intermittent 10-cm (2750 MHz) pulsed microwaves for 16 hours a day at 100, 500, or 2500 JlW/cm2 over a 4-month period; control animals were sham exposed. This microwave treatment significantly decreased the animals' motor activity (100 and 500 JlW/cm2), increased electrodermal sensitivity (500 and 2500 JlW/cm2), decreased the glycogen content in liver and brain (all intensities), decreased cytochrome oxidase activity in brain mitochondria (2500 JlW/cm2), and changed some immune indices (500 and 2500 JlW/cm2). Mathematical analysis of the collected data has predicted the "no-effect field intensity" to be at 10 JlW/cm2, which was therefore recommended as a maximum permissible irradiation level for residential areas.

The authors continued with studying the effects of a civil aviation radar system (Tomashevskaia and Dumanskii, 1988). Adult white rats were exposed to 400-Hz microwave pulses for 16 hours/day for 4 months using various carrier frequencies, intermittence regimens, and field intensities: (1) 2750 MHz, 50, 100, or 500 JlW/cm2, 40-Jlsec bursts, 1 burst per 20 sec, (2) 1310 MHz, 20 or 100 JlW/cm2, 28-Jlsec bursts, 1 burst per 10 sec, and (3) 850 MHz, 20, 100, or 500 JlW/cm2, 67-Jlsec bursts, 1 burst per 6 sec. The experiments established that exposures at 20 and 50 JlW/cm2 produced no statistically significant effects compared to sham-exposed controls. Exposures at 100 and 500 JlW/cm2 could cause statistically significant changes in the activity of several marker enzymes (cholinesterase, cytochrome oxidase, succinate dehydrogenase, and ceruloplasmin); the magnitude of changes in different tissues varied depending on the particular irradiation regimen.

Navakatikian and co-authors (1991) exposed rats to 3000-MHz pulsed radiation for 2 months, 12 hours/day during the nighttime. Bursts of 20, 4, or 2 pulses (2-Jlsec pulse width, 400-Hz repetition rate) were emitted every 20, 3.75, or 2.07 sec, respectively, simulating antenna rotation speeds of 3, 16, and 29 rpm. The average energy output was virtually the same for all the pUlsing regimens (60, 64, and 58 pulses/min). The incident power density levels studied were 0.1, 0.5, and 2.5 mW/cm2. Exposure effects evaluated were the animals' exploratory activity (measured as an amount of movement in a special maze) and conditioned avoidance reflex in a shuttle box test. Most of the exposure regimens slightly stimulated the locomotor activity (except 20-pulse bursts at 0.5 and 2.5 mW/cm 2). Performance in the shuttle-box after exposures was characterized by decreased latency of responses and increased number of correct responses. Overall, the changes were interpreted as a prolonged and moderately expressed activation of the central nervous system.

Bioeffects of the above-mentioned exposure regimens were explored further and compared to the effects of 2450 MHz, O.Dl-l mW/cm2 CW irradiation (Navakatikian and Tomashevskaya, 1994). In particular, it was found that 2-month CW exposure caused suppression of the central nervous system, in contrast to weak activation under pulsed irradiation. Histological examination of the thyroid gland revealed functional activation of the zona fasciculata after pulsed microwave exposure. Pulsed microwaves consistently and reproducibly decreased blood levels of insulin and testosterone, while CW exposure had no

281

effect. The authors have speculated that inhibition of behavior was a direct effect of irradiation on the nervous system, whereas activation might be mediated by hormonal changes.

Another study has replicated the exposure conditions of meteorological radar personnel (Navakatikian et al., 1991). Adult female rats were subjected to simultaneous irradiation from 9375- and 1765-MHz sources for 12 hours daily, 30 min each hour. The average incident power levels for 9375 MHz were 0.12, 0.24, and 0.36 mW/cm2; for 1765 MHz, the power output was set 24 times less, so the integral intensities used were 0.125, 0.25, and 0.375 mW/cm2• Modulation pulse width and repetition rate were 1.1 J.l.s, 475 Hz for 1765 MHz,

and 2 J.l.s, 300 Hz for 9375 MHz. For 9375-MHz microwaves only, the pulses were assembled into 20-msec bursts and delivered at a rate of 1 burst every 10 sec. Behavioral endpoints after 1, 2, 3, and 4 months of everyday exposure were compared with those in a sham-exposed group. The exploratory activity in animals was slightly but statistically significantly suppressed after I-month exposure at 0.375 mW/cm2 and after 2-months exposure at 0.125 mW/cm2• Conditioned reflex performance in a shuttle box was suppressed after 2 months of exposure at 0.25 mW/cm2. After 3 and 4 months of exposure, all behavioral parameters were the same as in the control group. Overall, behavioral effects observed in this study were regarded as "weak and unstable".

In a study by Grigoriev et al. (1995), adult rabbits were exposed to 1.5-GHz pulsed microwaves for 30 min a day for 1 month, excluding weekends and holidays (16-ms pulse width, 0.3 mW/cm2 peak power, 0.12-Hz pulse repetition rate). Exposures were performed in an anechoic chamber, one animal at a time. During the exposure, animals were placed into a special Perspex cage (40 x 40 x 40 cm) with a built-in piezoelectric probe for recording motor activity. The authors performed two identical series of experiments. In each series, 5 animals were exposed and 5 were sham-exposed. Sham exposures were done in the same chamber in a random sequence with microwave exposures. In both series, the motor activity of the sham-exposed animals was maximum in the first 2 days, reflecting an orientationexploratory reaction. The later period (until the 15th day) was characterized by extensive fluctuations of motor activity, which was interpreted as development of adaptation. After 2 weeks of treatment, motor activity stabilized at a significantly lower level, showing that adaptation is completed. In exposed animals, this decrease of motor activity was not significant, and this effect was observed in both series of experiments. For example, in the control group of the second series, the number of movements recorded during 30 min of

sham exposure decreased from 117 ± 13 (the first 2 weeks' average) to 64 ± 7.8 (the second 2 weeks' average, p<.05). The respective numbers for microwave-exposed animals were 120 ± 11 and 105 ± 13 (p>.05). This irradiation effect was regarded as inhibition of the adaptive reaction, or disadaptation. The 30-min dynamics of motor activity did not show any difference between exposed and control animals, but one type of movements (rapid moving of paws up and down) significantly increased in the exposed animals in the second 2 weeks of the experiment. This type of activity pointed to an enhancement of excitatory processes in the central nervous system, increased anxiety and distress.

Physiological effects of fields generated by ship radio transmitters (13 MHz, 250 or 500 VIm, modulated by Morse code) were reported in a series of studies by Minkina et al. (1985) and Nikitina et al. (l989a,b). Male albino rats (160-180 g body weight) were exposed to the field for 2 hours a day, 5 times a week, over a 2- or 6-week period. The authors did not supply information as to whether or not these exposures affected the body temperature of the

282

animals. After the course of exposures, animals were decapitated and various tissues were taken for morphological and histological examination.

In the first of these studies, 2 weeks of exposure at 500 Vim slightly activated the neurosecretory function of supraoptic and paraventricular nuclei in hypothalamus. These alterations disappeared after 6 weeks of exposure, but, by this time, the mass of the adrenals was considerably decreased. Irradiation at 250 Vim for 30 days reduced the blood plasma level of corticosterone, with no other effect on adrenals or hypothalamus. Changes observed in the thyroid gland morphology indicated activation of its function by 500 Vim exposure and suppression by 250 Vim exposure. Out of studied tissues and organs, spermaries proved to be the most sensitive to irradiation. Disorders of spermatogenesis, hemorrhagia, degenerative and dystrophic changes increased with longer course and higher intensity of exposures.

The second study (Nikitina et aI., 1989a) was primarily focused on cytogenetic effects of exposure (500 Vim for 10 days). Chromosome aberration frequency significantly increased in bone marrow cells (to 10-11% versus 5-6% in sham exposed controls, p<.05), but not in corneal cells. At the same time, the mitotic index in corneal cells fell from 1. 0 1 % to 0.66% (p<.05). The ascorbic acid concentration in the hypophysis and adrenal cortex increased from 264 ± 17 mg% and 403 ± 15 mgolo in control animals to 361 ± 27 mg% and

478 ± 24 mg%, respectively. The size of the adrenal cortex cells decreased. In thyroid gland, the growth of colloid accumulation in follicular cavities was observed. The follicular area

filled with colloid increased from 10.8 ± 2.8% to 29.7 ± 6.1% after irradiation. Irradiation disrupted spermatogenesis and reduced the mitotic activity of spermatic epithelium. The observed morphological changes indicated suppression of adrenocorticotropic hormone secretion and of the adrenal cortex and thyroid gland function. The production of luteotropic and follicle-stimulating hormones apparently was not affected.

In the third study (Nikitina et aI., 1989b), the authors suggested that individual features of the animals, namely, "the type of organization of their nervous function", might playa role in their sensitivity to radiofrequency exposure. As in the previous studies, the experiments were performed in male albino rats. Based on the animals' characteristic behavior in an openfield test, they were assigned to groups with "low-entropy" (LE) and "high-entropy" (HE) organization of the nervous function. After 60 days of exposure at 500 Vim, the authors evaluated the condition of the animals by 30 various indices (morphology of thyroid, adrenals, hypophysis; blood level of testosterone, luteinizing and follicle-stimulating hormones, etc.) and the condition of their progeny on the 20th day of embryo development by 13 other indices (body mass and length, organ pathology, ossification rate, etc.). Examination of the sham-exposed group revealed 7 indices that were significantly different in the LE and HE animals; in the exposed group, there were 17 such indices. The authors provided three examples that were characteristic for these differences. The morphological changes in the thyroid gland evidenced for its activation in both LE and HE exposed animals; however, in the LE specimens this actually was manifested as an increment in the epithelial cells height, and in the HE specimens it was an increment in the cell nucleus area. Exposure significantly diminished body mass in the exposed LE animals (by about 10% compared with LE controls), but did not change it in HE animals. There was no difference in the rate of ossification in the progeny of the control HE and LE animals, while the EMF treatment decreased this index in the offsprings of the LE specimens and increased it in the offsprings of the HE specimens. Thus, the experimental results revealed considerable differences in the EMF sensitivity between the LE and HE animals. It was recommended to consider such individual differences in future behavioral studies and in the development of safety standards.

283

Perhaps the most pronounced and unambiguously adverse effects of low-intensity microwave radiation were reported by Lokhmatova (1994). Effects ofa 4-month exposure for 2 hours/day (3-GHz, 0.25 mW/cm2) on reproductive organs were studied in 22 adult male rats. While the transmitter operated in a CW mode, an intermittent irradiation regimen was achieved by rotating the transmitter hom antenna at 22 rpm. A similar group of 22 animals underwent the same course of sham exposures. Morphological and histochemical analyses of the testes and epididymides were performed right after the end of the exposures and also 4 months later. The experiments revealed explicit morphofunctional disorders in RF-exposed animals. The number of normal seminiferous tubules decreased to 28.2 ± 5.3% (compared with 47.1 ± 5.3% in the controls), while the number of the tubules with dying cells increased to 27.2 ± 4.4% (14.7 ± 3.3% in the controls, p<.05). The number of Leydig cells markedly decreased, and some of them showed pyknotic and destructive alterations. RF exposures increased alkaline phosphatase activity in the vicinity of basal membranes and in the spermatic channel walls, and increased levels ofNADH and succinate dehydrogenases. Most of these disorders showed only a subtle or no trend to recovery in 4 months after the end of exposures. The author concluded that the changes evoked by the prolonged microwave exposure are likely to result in stable impairments of production and balance of steroid hormones and premature loss of reproductive function.

BIOEFFECTS OF EXTREMELY mGH POWER PULSES (EHPP)

Nowadays, EHPP technologies and applications are among the fastest growing areas, but very little is known about EHPP bioeffects and health hazards. The overall number of biological studies with RF pulses of 1-100 kW/g does not exceed 2-3 dozen (including meeting abstracts), and only a few isolated studies have explored still higher peak field intensities. Russian EHPP studies employed the most advanced pulsed power sources (which are still not available for biologists in the West), but the quality of reported biological research often is questionable.

Tambievet al. (1989) studied the effect of lO-ns wide, 200 kW/cm2 RF pulses (3-cm wavelength, 1 pulse per 6 min) on the growth rate and photosynthetic oxygen evolution in blue-green alga Spirulina platensis and unicellular green algae Platymonas viridis. It was found that exposure to 10-15 pulses stimulated cell culture growth and oxygen evolution. For example, by the 30th day after a single exposure, the biomass of S. platensis cultures increased by 52% in comparison with sham-exposed controls, and that of P. viridis increased by 17%. Maximum stimulatory effect (by up to 115%) was achieved when a treatment by lO-15 EHPP was performed immediately after a 30-min exposure to 2.2 mW/cm2, 7.I-mm wavelength CW radiation (this mm-wave irradiation alone enhanced the growth by about 50%). The same combined treatment caused the most pronounced enhancement of the photosynthetic oxygen evolution (about 1.5 times). Further experiments demonstrated that a greater number ofEHPP (20-25) had the opposite effect, i.e., suppressed the alga growth.

Deviatkov et al. (1994, 1998) studied the effects of 10-ns EHPP (3-cm wavelength, 100-MW peak output power) on malignant neoplasm development in vivo and in vitro. Exposure of Wistar rats for 9 days, 43 pulses/day, prior to inoculation with Walker carcinoma, decelerated the neoplasm growth rate 1.5 times and increased the life expectancy by 25-30%. In vitro exposure of Walker carcinoma cell suspensions (80 kV/cm, 6 pulses/min for 5 to 30 min) increased the number of degenerating cells and those in a stage of lysis. Further

284

experiments studied combined effects of EHPP and an anti-tumor drug endoksan on alveolar hepatic cancer RS-l in rats. The drug treatment and exposures (6 pulses/min, 30 min/day for 7 days) began when the tumor volume reached 3-4 cm3. By the 55th day after tumor inoculation, the tumor volume was 84.7 cm3 in controls, 14 cm3 in the endoksan-treated group, and 8.9 cm3 in the group treated by endoksan and EHPP. A similar experiment with Lewis lung carcinoma found that EHPP treatment (6 pulses/min, 30 min/day for 4 days and then for 5 days more, after a I-day interval) suppressed metastases, from 3.0 points (untreated controls) to 2.3 ± 0.4 (EHPP only) or 0.6 ± 0.2 (EHPP + endoksan); endoksan only

decreased this index to 1.4 ± 0.4. For a health risk assessment, 58 animals were exposed to

EHPP (from 130 to 720 pulses in 5 sessions) and compared with 64 control animals. Periodic examinations (3-4 times a month) over a period of 1-1.5 years have not revealed any modifications in behavioral responses or in the general state of health of the animals. Autopsy of animals in one year after the exposure did not reveal any pathological modifications in the liver, kidneys, adrenals, thymus, spleen, or other organs. We have to

note, however, that lack of sufficient detail on the experimental data, protocols, dosimetry, and statistics in this study makes it difficult to evaluate the validity of reported findings.

An interesting EHPP study has been recently reported from the Institute of High Current Electronics in Tomsk, Russia (Bolshakov et al., 1999). The study was designed to check out several likely mechanisms ofEHPP biological action, namely (1) membrane electroporation by high instant voltages of the electric field, (2) direct influence of the intense electric filed on macromolecular charged complexes, such as unfolded DNA sites and cytoskeleton fibers, and (3) changes in the rate of biochemical reactions, diffusion of substances and cell migration due to temperature gradients caused by EHPP absorption. These mechanisms were tested in experiments with (1) growth rate of Escherichia coli culture, (2) proliferation and growth of simple eukaryotic organisms (fungus Fusarium), and (3) ontogenetic development of Drosophila flies. All exposures employed a relativistic microwave generator with 200-MW output, 10-ns pulse duration, and 3-cm wavelength. The incident E-field in the exposure zone (about 25-cm diameter) was 1.5 MY/m, which corresponds to the incident power density of300 kW/cm2.

The growth of E. coli culture was measured from changes in its optical density. Two

identical thermostabilized (± 1 0c) vials with the bacterial culture were simultaneously placed in an anechoic chamber, but one of the vials (control) was shielded from microwave radiation. Exposures for 5 or 15 min at various EHPP repetition rates (up to 100 Hz) caused no significant changes in the cell growth rate. This result was interpreted as lack of cell electroporation by the EHPP pulses.

Proliferation rate of the Fusarium fungus was measured by micrometry, from the diameter of cell colonies and the length of hyphae. Exposure regimens were the same as in experiments with E. coli; in addition, some samples were exposed in cycles (5 min exposure, 5 min pause) for a total of 1 hour at the EHPP repetition rate of 6 Hz. All tested exposure regimens substantially decelerated the growth of hyphae, on the average from 0.338 ± 0.055

mmlhour (control) to 0.242 ± 0.033 mmlhour. This deceleration was equivalent to the effect of conventional heating by 20-25 °C; however, heating of samples during microwave exposure was less than 1°C.

For experiments with Drosophila eggs, larvae, and pupa, each group included at least 300 specimens. Each group was exposed or sham exposed just a single time, at different stages of ontogenesis. The exposure duration was 5 min or 60 min (12 cycles of 5-min exposure with 5-min intervals); the EHPP repetition rate was 6, 10, 16, or 22 Hz. All tested

285

exposure parameters typically caused a 2- to 10-fold increase of cases of interrupted specimen development. Exposures of l-hour-old embryos could cause developmental abnormalities (morphoses) in 6 to 9% of specimens (versus zero cases in parallel controls), increased imago lethality, and caused infertility in survivors. Additional experiments with imago flies established that 5 to 12% of flies died during a I-min exposure at EHPP repetition rates of 16, 50, and 100 Hz. At the rates of 6 and 10Hz, no flies died, but their first generation progeny became almost entirely nonviable (34% had morphoses, 38% died on the first day, and 100% were unable to lay eggs). Since the average heating by microwave exposures was negligible, the observed effects point to certain specific mechanisms of EHPP biological action.

SUMMARY

We conclude that RussianlFSU studies constitute an important source of information on pulsed RF bioeffects. Particular emphasis in these studies was given to RF-induced changes in the nervous system function, while such issues as RF-induced carcinogenesis apparently have not been a concern and were not studied at all.

While many (perhaps, most) of the studies were flawed, a number of good-quality studies have convincingly demonstrated significant bioeffects of pulsed microwaves. Modulation often was the factor that determined the biological response to irradiation, and reactions to pulsed and CW emissions at equal time-averaged intensities in many cases were substantially different. These results showed that bioeffects of pulsed RF may involve some specific mechanisms of interaction, which are not understood yet.

Most reported bioeffects of low-intensity pulsed microwaves were just subtle functional changes, which did not exceed the limits of normal physiological variation and could only be detected by sensitive physiological tests. However, some studies did report clearly pathogenic effects, and an independent confirmation of these findings would be of principle importance for understanding of the health hazards from RF exposure and development of safety standards. These studies deserve careful consideration and their replication in the West and/or with participation of Western scientists should be given a high priority.

ACKNOWLEDGEMENTS

The work was supported in part by the U.S. Army Medical Research and Materiel Command and the u.s. Air Force Research Laboratory (HEDR and AFOSR) under U.S. Army contract DAMDI7-94-C-4069 awarded to McKessonHBOC BioServices. The views expressed are those of the authors and should not be construed as reflecting the official policy or position of the Department of the Army, Department of the Air Force, or the United States Government.

REFERENCES

Afrikanova, L. A. and Grigoriev, Iu. G., 1996, Effects of various regimens of electromagnetic irradiation on cardiac activity (an experimental study), Radiats BioI Radioecol, 36/5: 691-699.

286

Akoev, I. G., Karanova, M. V., Kuznetsov, Y I., and Kolomytkin, O. Y, 1985, Action of a uhf field on GABA-ergic and acetyl cholinergic systems in synaptic transmission, Radiobiologiia, 25/3: 426-428.

Antimonii, G. D., Badikov, V. I., Kel, A A, Krasnov, E. A, and Sudakov, S. K, 1976, Changes in the self-stimulation reaction in rats under the action of a modulated electromagnetic field, Biull Eksp BioI Med, 82110: 1163-1165.

Arber, S. L. and Lin, J. C., 1984, Microwave enhancement of membrane conductance: effects of EDT A, caffeine and tetracaine, Physiol Chem Phys Med NMR, 16/6: 469-475.

Belokhvostov, AS., Osipovich, V. K, Veselova, O. M., and Kolodiazhnaia, Y A, 1995, Analysis of DNA ofretrotransposon-like genetic line-elements in blood plasma of rats exposed to radiofrequency electromagnetic waves, Radiats BioI Radioecol, 35/3: 356-363.

Bolshakov, M. A and Alexeev, S. I., 1992, Bursting responses of Lymnea neurons to microwave radiation, Bioelectromagnetics, 13/2: 119-129.

Bolshakov, M. A, Bugaev, S. P., Goncharik, A 0., Gunin, A Y, Evdokimov, E. Y, Klimov, A I., Korovin, S. D., Pegel', I. V., and Rostov, Y V, 2000, Effects of intense microwave radiation of nanosecond duration on some biological objects, Dokladi Akademii Nauk (in press, referenced by permission of the authors).

Chou, C-K, Yee, K-C., and Guy, A W., 1985, Auditory response in rats exposed to 2,450 MHz electromagnetic fields in a circularly polarized waveguide, Bioelectromagnetics, 6/3: 323-326.

Deviatkov, N. D., Betskii, O. Y, Kabisov, R. K, Morozova, N. B., Pletnev, S. D., Faikin, V. Y, and Chernov, Z. S., 1998, Effect of low-energy pulsed ehf and microwave radiation with nanosecond pulse duration and high peak power on biological structures (malignant neoplasms), Biomedicinskaya Radioelectronica, 1: 56-62.

Deviatkov, N. D., Pletnev, S. D., Chernov, Z. S., Faikin, Y V., Bernashevskii, G. A, and Shchitkov, KG., 1994, Effect of low-energy pulses of EHF and SHF-radiation of nanosecond duration with a high peak intensity on biological structures (malignant neoplasms), Dokl Akad Nauk, 336/6: 826-828.

Gapeev, A B., Yakushina, Y S., Chemeris, N. K, and Fesenko, E. E., 1997, Modulated extremely high frequency electromagnetic radiation of low intensity activates or inhibits respiratory burst in neutrophils depending on modulation frequency, Biofizika, 42/5: 1125-1134.

Gorbunova, A Y, Petrova, N. V., PortugaIov, Y V. and Sudakov, S. K, 1981, Acute experimental emotional stress in rabbits exposed to modulated electromagnetic fields, Izv Akad Nauk SSSR [BioI], 5: 774-780.

Grigoriev Yu. G., 1999, personal communication. Grigoriev, Iu. G. and Stepanov, V S., 1998, Forming of memory (imprinting) in chicks

after prior low-level exposure to electromagnetic fields, Radiats BioI Radioecol, 38/2: 223-231.

Grigoriev, Iu. G., Lukianova, S. N., Makarov, V. P., Rynskov, V. Y, and Moiseeva, N. V., 1995, Motor activity of rabbits in conditions of chronic low-intensity pulse microwave irradiation, Radiats BioI Radioecol, 35/1: 29-35.

Iurinskaia, M. M., Kuznetsov, Y I., Galeev, A L., and Kolomytkin, O. V., 1996, Reaction of the brain receptor system to the effect of low intensity microwaves, Biofizika, 4114: 859-865.

287

Kashtanov, S. I. and Sudakov, S. K., 1981, The predominant role of reticular fonnation of the midbrain in mediating the preventive effects of ultra-high frequency electromagnetic fields on excitation of emotiogenic hypothalamic centers, Biull Eksp BioI Med, 91111: 523-526.

Khitrov, Iu. A. and Kakushkina, M. L., 1990, The action of microwave radiation on potassium . ion transport and oxygen consumption in the perfused rat liver, Radiobiologiia, 30/2: 247-251.

Khramov, R. N., Sosunov, E. A., Koltun, S. V., llyaova, E. N., and Lednev, V. V., 1991, Millimeter-wave effects on electric activity of crayfish stretch receptors, Bioelectromagnetics, 12/4: 203-214.

Kim, Iu. A., Kasimbekov, I. K., Fomenko, B. S., Tigranian, R. E., 1986, Effect of pulse electromagnetic radiation on erythrocyte ghosts, BioI Nauki, 11: 37-41.

Kolomytkin, 0., Kuznetsov, V., Yurinska, M., Zharikov, S., and Zharikova, A., 1994, Response of brain receptor systems to microwave energy exposure, In: On the Nature of Electromagnetic Field Interactions with Biological Systems. A. H Frey, ed., R. G. Landes Co., Austin, TX, pp 194-206.

Konovalov, V. F. and Andreeva, L. A., 1989, Behavioral effects of VHF exposure, In: Influence of Electromagnetic Radiation on Biological Objects and Laser Medicine, V. I. llichev, ed., Far East Branch, Acad. of Sciences of the USSR, Vladivostok, Russia, pp 122-136.

Korotkov Iu. S., Burenkov, M. S., Burenkova, L. A., Pichugin V. Iu" Chunikhin, S. P., and Engovatov, V. v., 1996, The reaction of the tick Hyalomma asiaticum (Acarina, Ixodidae) to 1- to 4-GHz microwaves, Med Parazitol (Mosk), 4: 28-31.

Kuznetsov, V. I., Iurinskaia, M. M., Kolomytkin, O. V., and Akoev, I. G., 1991, Action of microwaves with different modulation frequencies and exposure times on GABA receptor concentration in the cerebral cortex of rats, Radiobiologiia, 3112: 257-260.

Lin, J. C., 1990, Auditory perception of pulsed microwave radiation, In: Biological Effects and Medical Applications of Electromagnetic Energy, O. P. Gandhi, ed., Prentice Hall, Englewood Cliffs, NJ, pp 277-318.

Lokhmatova, S. A., 1994, The effect of low-intensity prolonged impulse electromagnetic irradiation in the uhf range on the testes and the appendages of the testis in rats, Radiats BioI Radioecol, 34/2: 279-285.

Lukianova, S. N. and Moiseeva, N. V., 1998, Analysis of pulsed bioelectric activity of rabbit cerebral cortex in response to low-intensity microwave radiation, Radiats BioI Radioecol, 38/5: 763-768.

Lukianova, S. N., Rynskov, V. V., and Makarov, V. P., 1995, A comparison of conditioned avoidance reflex in rabbits fonned under the influence of permanent magnetic fields, ultra-high-frequency irradiation, light and sound, Radiats BioI Radioecol, 3511: 69-73.

Lukianova, S. N., Rynskov, V. V., and Makarov, V. P., 1995, Response of neurons of the sensomotor region of the cerebral cortex to low-intensity pulsed ultra-high frequency irradiation, Radiats BioI Radioecol, 35/1: 53-56.

McRee, D. I. and Wachtel, H, 1982, Pulse microwave effects on nerve vitality, Radiat Res, 9111: 212-218.

McRee, D. I., and Wachtel, H., 1980, The effects of microwave radiation on the vitality of isolated frog sciatic nerves, Radiat Res, 82/3: 536-546.

288

Minkina, N. A, Kuzminskaya, G. N., Nikitina, V. N., and Garina, I. A, 1985, Influence of the intermittent short-wave electromagnetic field on endocrine glands, Radiobiologiia, 25/6: 756-762.

Moiseeva, N. v., 1996, Experimental data on the reaction of brain neurons to low-intensity pulse-burst microwave irradiation, Radiats BioI Radioecol, 36/5: 710-713.

Navakatikian, M. A and Tomashevskaya, L. A, 1994, Phasic behavioral and endocrine effects of microwaves of nonthermal intensity, In: Biological Effects oj Electric and Magnetic Fields, vol. 1: Sources and Mechanisms, D. O. Carpenter and S. Ayrapetyan, eds., Academic Press, San Diego, pp 333-342.

Navakatikian, M. A, 1992, A method of studying of a defensive conditioned reflex of active avoidance, Zh Vyssh Nerv Deiat 1m I P Pavlova, 42/4: 812-818.

Navakatikian, M. A, Nikitina, N. G., and Zotov, S. V., 1991, Behavioral effects of the chronic combined exposure to 9375 and 1765 MHz microwaves, Gig Sanit 8: 52-53.

Navakatikian, M. A, Soldatchenkov, V. N., Bitkin, S. V., and Zotov, S. v., 1991, The status of the higher nervous activity in animals exposed to microwaves in conditions simulating the intermittent work of radio locators, Gig Sanit 7: 57-61.

Navakatikyan, M. A, 1993, A problem of standardizing methods for biological evaluation of an electromagnetic field effect in hygienic regulation, In: Hygiene oj Physical Factors oj Environmental and Industrial Surroundings, First International Congress (16-18 November 1993, Kiev, Ukraine), pp 68-69.

Nicitina V. N., Matskevich O. A, Dubrovina V. v., Minkina N. A, and Kuzminskaya G. N., 1989a, Cytogenetic and endocrinologic changes in experimental animals exposed to high-frequency electromagnetic fields, Gig Sanit, 5: 85-86.

Nikitina, Y. N., Suvorov, N. B., Minkina, N. A, and Shaposhnikov, E. S., 1989b, Peculiarities of the response of animals of various typological groups to electromagnetic radiation of high and very high frequency, Radiobiologiia, 29/5: 676-679.

Pakhomov, A G., 1993, Non-thermal microwave effect on nerve fiber function, Biofizika, 8/2: 367-371.

Pakhomov, A G., Dubovick, B. v., Degtyariov, I. G., and Pronkevich, AN., 1995a, Microwave influence on the isolated heart function: I. Effect of modulation, Bioelectromagnetics, 16/4: 241-249.

Pakhomov, A G., Dubovick, B. v., Degtyariov, I. G., and Pronkevich, AN., 1995b, Microwave influence on the isolated heart function: II. Combined effect of radiation and some drugs, Bioelectromagnetics, 16/4: 250-254.

Pakhomov, A G., Dubovick, B. V., Kolupayev, V. E., and Pronkevich, AN., 1993a, Effect of high-peak-power microwave pulses on isolated nerve function, Electro Magnetobiol, 12/1: 1-15.

Pakhomov, A G., Dubovick, B. v., Kolupayev, V. E., and Pronkevich, AN., 1992, Microwave-induced changes in nerve conduction: effect of modulation, In: Proceedings oj the J 4th Int. Con! oj the IEEE Engineering in Medicine and Biology SOciety (paris, France Oct. 29-Nov. 1. 1992), part 1, pp. 293-294.

Pakhomov, A G., Dubovick, B. v., Kolupayev, V. E., and Pronkevich, AN., 1991, Absence of non-thermal microwave effects on the function of giant nerve fibers, J Bioelectr, 10/1&2: 185-203.

289

Pashovskina, M. S. and Akoev, 1 G., 1996, Effects of 2375 MHz pulse-modulated microwave radiation on ATPase activity of the rat muscle actomyosin, Radiats BioI Radioecol, 36/5: 700-705.

Philippova, T. M., Novoselov, V. 1, Bystrova, M. F., and Alexeev, S. I., 1988, Microwave effect on camphor binding to rat olfactory epithelium, Bioelectromagnetics, 9/4: 347-354.

Popov, V 1, Novoselov, V. 1, Filippova, T. M., Khutsian, S. S., and Fesenko, E. E., 1996, Electron microscopy analysis of the effect of a modulated microwave radiation on isolated rat olfactory mucosa, Biofizika, 4114: 8S1-858.

Rynskov, V. V., Polyntsev, Iu. V, Lukianova, S. N., Makarov, V P., and Manas'eva, E. A, 1995, Evaluation of changes in electrophysiological and hormonal parameters in rabbits resulting from short-term low-intensity ultra-high-frequency irradiation, Radiats BioI Radioecol, 3511: 66-68.

Semin, Iu. A, Shvartsburg, L. K, and Dubovik, B. V., 1995, Changes in the secondary structure of DNA under the influence of an external low-intensity electromagnetic field, Radiats BioI Radioecol, 35/1: 36-41.

Sudakov, K V and Antimonii, G. D., 1977, The hypnotic effect of modulated electromagnetic field, Biull Eksp BioI Med, 34/8: 146-149.

Sudakov, K V., 1976, Modulated electromagnetic field as a factor of selective effect on the mechanisms of purposeful behavior in animals, Zh Vyssh Nerv Deiat, 26/5: 899-908.

Sudakov, K V, 1998, Action of modulated electromagnetic fields on the emotional component of the systems organization of behavioral acts in rats, Neurosci Behav Physiol, 28/6: 686-693.

Tambiev, A Kh., Kirikova, N. N., Lapshin, O. M., Betzkii, O. V., Novskova, T. A, Nechaev, V M., and Petrov, 1. Yu., 1989, The combined effect of exposure to EMF of millimeter and centimeter wavelength ranges on productivity of microalgae, In: Millimeter Waves in Medicine and Biology, N. D. Devyatkov, ed., Radioelectronica, Moscow, pp 183-188.

Tinney, C. E., Lords, J. L., and Durney, C. H., 1976, Rate effects in isolated turtle hearts induced by microwave irradiation, IEEE Trans Microwave Theory Tech, 2411: 18-24.

Tomashevskaia, L. A and Dumanskii, Iu. D., 1988, Hygienic evaluation of the biological effect of pulsed electromagnetic fields of 850-2750 MHz, Gig Sanit, 9: 22-24.

Tomashevskaia, L. A and Solenyi, E. A, 1986, Biological action and hygienic significance of the electromagnetic field created by coastal radar facilities, Gig Sanit, 7: 34-36.

Tygranyan, R. E., 1986, On the mechanism of pulsed microwaves effects on excitable structures, Vopr Kurortol Fizioter Lech Fiz Kult, 6: 11-14.

Zakharova, N. M., 1995, Enhancement of rhythmic processes in slices of the cerebral cortex under the action of pulse-modulated microwave radiation, Biofizika, 40/3: 695-696.

Zakharova, N. M., Alexeev, S. I., and Zhadin, M. N., 1993, Microwave effect on spontaneous electrical activity in slices of the cerebral cortex in vitro, Biofizika, 38/3: 520-523.

Zakharova, N. M., Karpuk, N. N., and Zhadin, M. N., 1996, Cross-correlation analysis of interconnections in neuron firing in neocortex slices under the effect of microwave irradiation. Biofizika, 4114: 913-915.

290

CONTRIBUTORS

Wei Chen, Ph.D. Departments ofDennatology, and of Physiology and Biophysics School of Medicine The University oflllinois at Chicago Chicago, IL 60612

John O. de Lorge, Ph.D. McKesson HBOC BioServices U.S. Army Medical Research Detachment Microwave Bioeffects Branch Brooks Air Force Base San Antonio, TX 78235

Bin He, Ph.D., EECS and Bioengineering Departments University ofIllinois at Chicago 851 S. Morgan Street, MC 154 Chicago, IL 60607-7053

Norio Iriguchi, Ph.D. Department of Biomedical Engineering Graduate School of Medicine University of Tokyo 7-3-1 Hongo, Bunkyo-ku Tokyo 113-0033, Japan

James C. Lin, Ph.D. EECS and Bioengineering Departments, University ofIllinois at Chicago 851 S. Morgan Street, MC 154 Chicago, IL 60607-7053

Shin-Tsu Lu, Ph.D. McKesson HBOC BioServices U.S. Army Medical Research Detachment Microwave Bioeffects Branch Brooks Air Force Base San Antonio, TX 78235

Andrei G. Pakhomov, Ph.D. Directed Energy Bioeffects Division Human Effectiveness Directorate Air Force Research Laboratory Brooks Air Force Base San Antonio, TX 78235

Michael R. Murphy, Ph.D. Directed Energy Bioeffects Division Human Effectiveness Directorate Air Force Research Laboratory Brooks Air Force Base San Antonio, Texas 78235

Berti! R. R. Persson, Ph.D. Radiation Physics Department University of Lund S-221 85 Lund, Sweden

Shoogo Ueno, Ph.D. Department of Biomedical Engineering Graduate School of Medicine University of Tokyo 7-3-1 Hongo, Bunkyo-ku Tokyo 113-0033, Japan

291

Dongsheng Wu, Ph.D. Bioengineering Department University of Illinois at Chicago, 851 S. Morgan Street, MC 154 Chicago, IL 60607-7053

292

Dezhong Yao, Ph.d. Bioengineering Department University of Illinois at Chicago, 851 S. Morgan Street, MC 154 Chicago, IL 60607-7053

Present address: Department of Automati~n & Biomedical Engineering Program, University of Electronic Science and Technology of China Chengdu 610054, Sichuan Province P.R. China

INDEX

Absorption threshold 210 Acetylcholine 250

receptors I S4 Acetylcholinesterase 277 Acoustic cue 219 Acoustic effect 212,218 Acoustic pulse 214 Acoustic signal 215 Action potential 148-149, 154-156,

160-161,174,227,241,270-271 Active transporters 153 Actomyosin 268 Acute exposure 273-280 Acute stroke 57-58 Adrenal cortex 283 Adrenocorticotropic honnone 283 Algorithms

cortical imaging 99-102, 105 CRES097 EEG deconvolution imaging 88 Laplacian weighted 91 noise convariance weighted 91 SVD 88-90, 99, 101, III weighted 90-92

Aminomethynol 268 Ampere's law 4 Analgesia 248 Angiogram 56 Anima1 experimentation 208 Anisotropy 57, 107 Annoyance 215 ANSIC95.1 standard 13 Anticancer drugs 129, 143 Anti-tumor drug 285 Aplysia pacemakers 226-227 Apomorphine 279 Arrhenius equation 125 Arterial blood pressure 222-223, 249 Arterial spin labeling 57 Ascorbic acid 283 Asthenic syndrome 267 Asymmetrical electroporation 156 Asymmebicaltransportl56 ATP hydrolysis 189-190 ATP synthesis 189-190 ATPase activity 268 Atropine 271-272 Auditory etTect 213, 215, 250 Auditory threshold 215 Avogadro's number 41 Avoidance behavior 243 Avoidance reflex 279, 281 Axonemes 273

Behavioral etTects 217-219 Behavioral paradigms 245 Bilayer lipid membrane 122 Bioelecbic source 74, 76, 81 Bioelectromagnetics 265 Biological effects 209, 216, 237, 242,

245 pulsed RF radiation 265-286

Biological systems 2 etTects ofEMFs on 2

ELF 18,35 RF radiation 23-35

electromagnetic source fields on 5-7 coupling mechanisms 14-18,23,

26 energy absorption perspectives

13-17 frequency dependence 14 induced fields 13 relaxation mechanism 11-12 studies on interactions 5-7,22

specific absorption rate studies 13-15,23-24,31,33

telecommwrication devices on 30-34 tissue electromagnetic properties of

4-5,24 Blair expression 241 Bleomycin 130-131, 138 Blood chemistry 228-231 Blood-brain barrier 224-226 Boron neutron capture therapy (BNCT)

137-138 Bound charge 11-12 Boundary condition 4, 83-85, 106, 109 Boundary element method 98, 105,

107 Bournlea pulse generator 248 Bradycardia 223, 271 Brain elecbical activity

imaging techniques 88-113 modeling of77-88 (See also Head

model) overview of73-74

Brain enzyme denaturation 209, 250 Brain imaging 54-59 (See also MRI) Brain model 16 Brain receptors 277 Brain surface potential 109 Brain temperature 210, 225, 233 Brownian motion 57,183 Bulk elecbical conductivity 13-14 Bulk heating 250 Buspirone 212

293

294

Ca2+ ATPase 150, 155, 184, 190 Caffeine antagonist 272 Cancer research 131 Capacitance 74-75,124-125,162-163,

171,241 Carboplatin 130 Carcinogenesis 284-286 Carcinoma 284-285 Cardiac IIJThytiunia 148, 159 Cardiac pacemakers 242-243 Cardiotoxic effectiveness 271 Cardiovascular effects 220-224, 249 CaJTier frequency 208, 223, 237, 269-

270 CaJTierless RF pulses 208, 237, 250-

251 absorption of 23 7 biological effects 237 mechanisms for tissue damage 239

Cation pumping 188-189 Cell fusion 159 Cell growth rate 284-285 Cell membrane 121-123, 148-150,

159-172,180,184,186-188,195-196 capacitance 74-75, 124-125, 162-

163, 171,241 directionality of 156 protection effects 156 study of electric field effects on 161

traditional methods 161 voltage clamp methods 163

Cell necrosis 160 Cell survival 126 Cell-attached patch-clamp method 163 Cellular telephones 30, 34 Cerebral cortex 276 Cerebral infarction 57 Cerebrospinal fluid 63-64 Ceruloplasmin 281 Channel currents 167, 177-178

voltage-gated 167, 169 Channel gating system 154, 180, 182 Channel opening 157-158,239 Channel open-threshold shift 178 Channel proteins 168

K+ 170 voltage-gated 169

Channel-open threshold 179-181 Charge injection/relaxation method

162 Charge layer 81 Charge movement currents 174-178,

182,192 Chemotherapy 143 Cholinesterase 281

Chromosomal aberrations 244 Chromosome aberration frequency 283 Chromosome abnormalities 231 Chronic exposure 280-284 Chronotropic effects 220 Circadian rhythmicity 236 Clonidine 212 Cober pulse 237 Cochlea 214-215 6OCo-gamma radiation 132-133 Cognitive behavior 237 Cole-Cole model139 Colloid accumulation 283 Color discrimination task 236 Compound action potential 227, 270-

271 Concentric three-sphere inbomogenous

head model 100, 110 Conditioned reflex response 279-282 Conductance 156-159, 161-163, 166,

179-180,182, 188 Conductivity 3-6,13,20,62,76,81,

106, 122-124, 135, 139, 142, 159 Conformational change 160-161, 170 Conformational damage 150, 153 Conformational oscillations of

enzymes 216 Conformational stsbility 188 Conformational stste 184 Conformational transition 148 Constrained inverse method 94 Continuous wave (CW) emissions 207,

216,219,265,271-273,277,281, 284

Convulsions 210, 247 Cortex

adrenal 283 cerebral 276 occipito-parietal 276 sensory-motor 228, 275-277

Cortical imaging 97-98, 105 Cortical imaging technique (Crr)

algorithm 99-102, 105 Cortical potential 98, 106-107, 11 0 Cortical potential imaging 74, 100-103

approaches to 98-111 Cortical potential map 87, 105, 110 CRESO algorithm 97 Critical duration 251 (See also

Exposure duration) Cross-linking reagent 161 Current density 20, 75-78, 113 Current rectification 186-187 Current-voltage relationship 162-163 Cytochrome oxidase 281 Cytotoxic drugs 129

d-amphetamine 212 Death 160 Deblurring method 98, 106-107 Defibrillation 148-149, 156, 159-160,

166 Defonnation energy 125 Degenerative disorders 283 Deoxy-hemoglobin (Deoxy-Hb) 57-59 Depolarization 160, 166, 197,269 Desipramine 212 Diazepam 212 Dielectric breakdown 121-124, 139 Dielectric constant 4-6, 32 Dielectric constant change 148 Dielectric pennittivity 10, 16, 18, 20 Dielectric resonance 61-62, 66-67 Diethylenetriamine-pentaacetic acid

133-137 Differential reinforcement 217 Diffusion imaging 40, 57, 67 Dihydropuridine receptors 154, 174 Dimensionless ratio 3 Dipole 79, 99, 121 Dipole antenna 5 Dipole field 79 Dipole layer 99-\03, \05, III Dipole localization method 73, 97 Dipole moment 12, 79, 99, \05, 170,

183 Dipole potential 80 Dipole source 76 Dipole source model 79 Dipole-layer imaging 112 Dipole-layer model 81 Discrepancy technique 97 Discrete Picard condition 96 Discrimination cue 218 DNA stability 268, 274-275 Double-layer model 81 Drug-induced convulsions 247 Duty factor 209, 217, 223-224,230,

232,234,241

Eccentricity 102-103 Echo signal 48 Echo time 43-44, 54 Echo-planar imaging (EPI) 44, 57 Ecosystem 273 Eddy current 21-22,33,36,61-62,66 Edema 128, 148 EEG deconvolution imaging 88

algorithms for 88 research activities in 74

EEG inverse imaging 98-99 brain electrical activity by 97-98

Effects

acoustic 212, 218 auditory 213,215,250 behavioral 217-219 biological 209, 216, 237, 242, 245 cardiovascular 220-224, 248-249 chronotropic 220 mutagenic 246 nervous system 226-228 ocular 219-220

Electoconfonnational change 170,174 Electric dipole moment 10 Electric field 10-14, 18, 121, 124-125,

134, 149-150, 159-165, 187-188, 190-191,195-196,237,239,247-248,285 response of boumd charge II

Electric field pulse techniques (See High voltage impulse treatment)

Electric field-induced dysfunction 160 Electric flux density 12 Electrical injury 147-148, 152, 155,

160, 163, 169, 174, 182, 197 Electric potential 73, 76-77, 81 Electrical neutrality 79, 83 Electrochemical potential 149 Electrochemotherapy 130, 139 Electroconfonnational coupling (ECC)

150, 183, 187 Electroconfonnational damage 159,

169-170,182,196-197 Electroconfonnational energy 183 Electro-coupling confonnational

change theory 188 Electrodennal sensitivity 281 Electroencephalogram (EEG) 73 Electrogenic pump 149, 154-155, 188,

191,196 Electromagnetic energy 1, 5, 35, 213,

243 spatial distribution of 5

Electromagnetic pulse (EMP) 208, 215,242 studies on biological effects 242

Electromagnetic radiation 2 Electromagnetic source fields 5-7 Electromagnetic theory 4, 10 Electromagnetic wave 4, 10 Electro-medicated leakage 161 Electromotive force 40-41 Electro-osmosis 156 Electropermeabilization 121, 123, 126,

134-139,143,163 Electroporation 121, 129-130, 143,

156-160, 166, 182,239,240 Electroporation threshold 169 Electro-pulsation 137

295

296

Electro-radio-therapy 143 Electrostatic dipole 8 Electrostatic field 8 Elementary dipole antenna 5, 7-10, 35-

36 energy characteristics of 9 fields of7-10

ELF field II, 35 EMF Database 266 Endoksan 285 Energy absorption 13-17, 136, 149,

183,210,237 Energy coupling II Energy deposition 24, 209 Energy transport 9, 196 Enhanced radiation sensitivity 143 Entropy 283 Enzyme denaturation 209-210 Enzyme dynamics 184 Epicortical field 108 EPISTAR57 E-polarization 29 Equilibrium potential 151 Equivalent dipole source density 76 Equivalent electrical circuit model

122-126 Equivalent monopole source density 76 Equivalent surface-source imaging III

concept of 112 Event related potential 75 Evoked potentials 75, 228 Excitation I Excitation pulse 44, 57, 63 Excitation-contraction coupling 154,

174 Exposure duration 211, 213, 215-216,

218,220,240,243,245,251 Extremely high power pulses (EHPP)

284 bioeffects of284-286

mechanisms of 285

FoFJ proton ATPase 155 FoFJ-ATPase 188 Far field zone 5, 7-8, 34-36 Faraday's law 4 Fertility 244 Fibroblast cells 126 Field echo (FE) imaging 44, 48, 50, 53 Field echo (FE) signal 46, 48, 57 Finite element model 98, 105-106 Finite-difference time-domain (FOlD)

technique 238 FLASH sequence 56, 65 Flip angle 64-67 Fluorescence 267

Fluorescent probe 267 Flux density 3 Force balance equation II Former Soviet Union (FSU) 265 Fourier transformation 12, 46-53 Free induction decay 42 Free space 3, 5, 12, 36 Frequency dependence 5, 14, 185, 189-

190 Frequency

carrier 208,223,237,269-270 chromosome aberration 283 dependence 5, 14, 185, 189-190 encoding 50 Lamor 61-63, 67 modulation 267-269, 272, 279 phase plane 49 proton resonance 46, 50 resonance 270 spin resonance 44

Frequency-encoding 50 Frequency-phase plane 49 Functional neuroimaging (tMRI) 40,

53,57-59

GABA receptors 277-278 Gamma-camera imaging 133-134 Gating charged particles 178, 180-181 Gauss electric law 4 Gaussian excitation 44 Gaussian pulse 238 Gaussian white noise 10 1-102 Gene therapy 121, 138, 143 Genetransfectionl43,159 General inverse 88-89 Genome stability 275 Glutamate binding 278 Glutamate receptors 277 Glycogen 281 Goldman-Hodgkin-Katz (GHK)

equation 152 Gray matter 54 Green's identity 108-109 Green's theori:m 76 Gypsy pulse 237

Hannonic continuation problem 100 ~-camphor binding 272 Head geometry 103 Head model 74, 87-88, 100-104, 107-

108, 110,216 Head volwne conductor model 81, 98,

lll, 113 Health risk assessment 285 Hematological disturbance 244-245 Hematopoiesis 229

Hemopoiesis 228 Hemorrhagia 283 Hepatic cancer 285 ~-glutamate 277 High energy deposition 209 High peak power radio frequency

pulses biological effects of207-251

auditory 213-215 behavioral217-219 blood chemistry 228-231 blood-brain barrier 224-226 brain studies 233 cardiovascular 220-224 carrierless pulses 237-245 electromagnetic pulses 242 Iymphoblastoid transformation

231-233 microwave evoked body

movements 210-212 nervous system 226-228 ocular 219-220 overview of 207-208 single pulse 209 startle reflex 212-213 TEMPO experiments 234-237 thermal sensation 213 ultra-wi de-band pulses 245-249

High voltage impulse treatment animal studies 129-130 clinical trials 131 combination with BNCT 137-138 combination with cytotoxic drugs

129-131,137 combination with radiation therapy

J3\ combination with radionuclide

therapy 131,133,137 combination with

radiopharmaceuticals 133 DNA delivery in gene therapy 138-

139 effect of duration in 136 effect of electric field strength 134-

135 factors affecting control of 139 in vitro treatment of fibroblast cells

126-127 in vivo effect of 128-129 methods for optimization of 139 modeling approaches to 139-142 molecular and cellular effects 126,

139 overview of 121-126 tumor studies in rats 128-133

High-affinity choline uptake 228

Hindenburg-2 (H-2) device 246 ~-muscimoI277 Hodgkin-Huxley nonlinear membrane

model 241 Homeostasis 240 Homogenous head model 107 Hormonal effects 283 Horseradish peroxidase 224-226 Hot spots 27, 34 H-polarization 29 Human body

coupling of electromagnetic energy to 16

Hydroperoxide 126 Hydrophone 215 Hyperactivity 248 Hyperpolarization 268 Hyperthermia 223, 226, 230-231 Hypophysis 283 Hypotension 222, 250 Hypothalamus 283

ICNIRP 1998 208 IEEE 1999 208, 237 hnage contrast 53-55 hnaging techniques (See also MRI)

brain electrical activity 54-59, 88-113 cortical 97-102, 105 cortical potential 74, 100-103 diffusion 40, 57, 67 dipole-layer 112 echo-planar 44, 57 EEG deconvolution gg EEG inverse 98-99

equivalent surface-source III field echo 44, 48, 50, 53 functional neuroimaging, 40, 53,

57-59 ganuna-camera 133-13 inverse III spatial deconvolution 74, 90, 99 spin-echo 48-50, 53,63 spin-warp 44-55 turbo gradient spin echo 54-55 turbo spin echo 57

hnmunocompetence 231 hnpedance 7-8, 14,24,30,36, 139-

142, 148 role in brain imaging 59-60, 67

hnpedance spectrometry 139, 143 hnprinting behavior 279 Individual change index 280 Induced electric field 8, 10, 13, 33 Inductance 75 Infertility 286

297

298

Infrared radiation I Inhomogenous head model 74, 103-

104, 108 Insulin 281 Interfacial polarization 124-125 Interstitial potential 160 Intrinsic impedance 7-8, 14,24,30,36 Inverse imaging III

algorithms 74 Inverse problem 88 Inversion recovery 54 Ion channel 148, 153-154, 158, 177,

241 currents 163 functional reduction 166 opening 156, 159 proteins 166, 178

Ion flux 196 Ion leakage 148, 161 Ion permeability 154 Ionization I Iron kinetics 228 Isotropic media 12 I-Vrelationship 186-187, 193-194

Joule heating effect 148,153,169-174, 182, 190

Joule's Law 170

K+ channel 178, 180, 197 conductance 178 currents 167-173 functions 173, 177 proteins 174, 178, 180 shock effects 168

Kentech PBG3 device 247 K-polarization 29 k-space 50-57

Lamor frequency 61-63, 67 Laplace's equation 18, 83 Laplacian weighted algorithm 91 Layer resonance 25 L-curve method 96 Leakage 161 Leakage current 156-157, 165-169,

172-173,193 Legendre function 82 Leukemia 243-244 Leukocytosis 229 Lewis lung carcinoma 285 Leydig cells 284 Ligand-gated channel 154 Ligand-mediated transport 182-183 Lipid peroxidation 126 Locomotor activity 248, 281-282

Lorentz curves 185-186 Low-field activation 150 Luminescence 126 Lymphoblastoid transformation 231-

233 Lymphocytosis 229-230

Magnetic field 4,10,14,21,244,280 Magnetic flux 4 Magnetic flux density 41-42 Magnetic momentum 57 Magnetic resonance angiography

(MRA) 40, 56-57 Magnetic resonance imaging (MRI)

31, 39 (See also Imaging techniques) applications of 56-58 brain structure by 54-60 experimental studies 62-67 modeling of 31 overview of 39-40,44, 67 principles of 61-62 slice selection 44-46, 48, 50

Magnetic resonance spectroscopy (MRS) 59

Magnetization 40, 42 Marker enzymes 281 Mass density 13-14 Maximum intensity projection 56 Maxwell equations 2-5 Membrane breakdown 126 Membrane capacitance 5 Membrane depolarization 149 Membrane electroporation 148, 153,

159-161,165,169-171,240,285 potential threshold for 166, 168

Membrane electroporation current direct observation of 165

Membrane electroporation threshold 177

Membrane enzymes 150 Membrane holding potential 195 Membrane permeability 210 Membrane peroxides 126 Membrane potential 124-125, 150,

155-164,170,172,175-183,186-188,192- 195,197,240 comparison of 177 difference 150 field-induced 152 intrinsic ISO role of membrane structures on 150 threshold 169, 177-178

Membrane proteins 147-148, 153-155, 159-162,182,184,188,197165, 169-170,173

activation by electric field 182 Membrane resting potential 148-149,

152-159,186-187,194-195,197, 241 depolarization of 166

Membrane transport system 183 Memory consolidation 245 Metal electrodes 279 Michaelis-Menten kinetics 184 Microwave effect 270-273, 278-279 Microwave evoked body movements

210-212,250-251 Microwave hearing 213, 250 Microwave pulse 207, 210-212, 214-

218,220-221,249,268-274,277, 281

Minimum norm 89, 92-93, 95, 99 Minimum product method 96 Mitomycin C 130 Mitotic index 229, 231, 283 Mobile counter ion 5 Models

boundary element 98,105, 107 brain 16 brain electrical activity 77-88 Cole-Cole 139 concentric three-sphere inhomogenous head 74,

100,103-104, 108, 110 dipole layer 81 dipole source 79 double-layer 81 equivalent electrical circuit 122-

126 fmite element 98, 105-106 head 74,87-88, 100-104, 107-

108, 110,216 head volume conductor 81,98,

1Il, 113 high voltage impulse treatment

139-142 Hodgkin-Huxley nonlinear membrane 241 homogenous head 107 human head tank 108 monopole source 77 MRI31 oscillatory barrier 188 parallel-conductance 122-123,

139 RC-circuit 139 RF energy coupling 28, 30 SENN241 single layer surface-source 112 single-layer 81 source 77, 79, 81, 88, 105, 112

source-conductor 100 spherical shell 19, 240 surface compartmental 150, 188 surface-source 112 three-shell realistic geometry

head 110 three-shell volume conductor

109 three-sphere conductor 81 three-sphere head 88, 10 1 volume conductor 105 volume source 76

Modulation 208,251,271-274,277-279,282

Modulation frequency 267-269, 272, 279

Molecular pump 197 Monopole 77 Monopole source model 77 Moore-Penrose inverse 90 Morphine 212, 248 Morphological effects 283-284 Morphoses 286 Mortality 247 Mortality ratio 243 Motility index 268-269 Motor stereotypy 279 Mutagenesis 246 Mutagenic effects 246

Na+/K.+ ATPase 149-150, 155, 161, 184, 186-188, 190, 194, 196

Na+/K.+ pump current 191, 193-195 asymmetric characteristics of 194 measurement of 195

NADH233 Naloxone 212 Na+ channels 177 Natural ecosystem 273 Near field zone 5,8-9, 14,29,33-36 Nerve vitality 271 Nervous system effects 226-228 Nervous system function 283 Nervous system suppression 281 Neurons 210, 272-276 Neurosecretory function 283 Neurotoxins 167 Neurotransmitters 209 Nitric oxide synthase 248 NMR signal 42 Nociception 248 No-effect field intensity 281 Noise convariance weighted algorithm

91 Noise-induced phase transitions 190

299

300

Non-invasive method 39, 73, 77,197, 249

Non-selective permeability 159 Nuclear magnetic resonance (NMR) 39

Objective function 94 Occipita-parietal cortex 276 Ocular effects 219-220 Ohm's law 78 Orientation behavior 244 Orthogonal projection temporal

constraint method 94 Oscillating electric field 155, 182-186,

194,196-197 Oscillating membrane potential 195-

196 Oscillation perturbation 149 Oscillatory activation barrier model

150 Oscillatory barrier model 188 Ouabain 161, 192-193, 196

Parallel-conductance model 122-123, 139

Parasympathetic blocker 271 p-choloramphetamine 212 Penetration 36 Penetrationn depth 25 Perfusion 40, 57 Permeability 3-5,10,156,159,180,

224,226 Permeabilization 156-157 Permittivity 3-6, 10, 12, 32, 67 Phase transition temperature 233 Phased-contrast (PC) method 56 Phase-encoding 48, 50 Phenanthroline 161 Phospholipid bilayer 148, 152-153,

157-158, 160-162, 165, 169, 182 Photosynthetic oxygen evolution 284 Physiological effects 279-282 Pilocarpine 212 Pimoxine 212 Plane wave 8, 10, 14; 24, 29 Point current source 81-82,86, 113 Point source 76-77 Point-source layer 81 Poisson's equation 75, 98, 106, III,

160 Polar molecule 5 Polarity 171, 173, 197 Polarization 12 Polarization dependence 29 Poloxamar 148 Porphyrin 137 Porsolt swim test 236, 245

Potential threshold 166, 168 Power density 9, 14,24,213,215,

217-219,221,224-225,234,236-237,269,277-278

Power deposition 14,23,32 Poynting vector 8 Pressure wave 215 Principle of superposition 78-81 Propranolol 271-272 Protein damage 174 Protein pump 184 Proton density 42 Proton resonance frequency 46, 50 Pulse duration 208,210-211 Pulsed radiofrequency radiation

biological effects of 265-286 overview of 265 research studies in RussiaIFSU

266-286 animal 273-284 in vitro 267 in vivo 267

Quadrature demoduration 46, 50 Quasi-static approximation 75 Quasi-static fields 10

Radar systems 245, 249, 280, 282 Radar transmitters 207 Radiation I (See also RF radiation)

dosimetry 13 electromagnetic 2 sensitivity 143 therapy 131,143 zone 5, 8,29

Radiative region 9, 35 Radio transmitters 282 Radioactive staining method 161 Radiofrequency disease 267 Radiofrequency in medicine I Radioinununo-therapy 143 Radiolabelled agonists 277 Radionuclide therapy 131, 134, 137,

143 Radiopharmaceuticals 133 RC-circuit model 139 Reactive oxygen species 270 Reactive region 9, 35 Reading-out 46-50, 63-66 Recovery time course 166 Reflection 24 Reflection coefficient 24 Refocusing pulse 44 Regularization 95 Regularization method 95 Regularized inversion 98

Relaxation 40, 42-44, 53, 140, 183 mechanism of 11-12

Remission 128, 133 Repetition time 54, 63 Reproductive capability 244, 284 Resistance 123, 162, 164,227 Resistivity 75, 107, 152 Resonance frequency 270 Resonance peak 43 Resonance transduction 150 Resonant coupling 29 Respiratory burst 269-270 Resting electrical potential 122 RF energy coupling 27

bodies with curvature effects 26 modeling 28, 30 near zone 29 polarization dependence 29 role of orientation 29

RF radiation 11,23, 35 biological effects 33 (See also

Biological effects) coupling mechanisms for 23-35 modeling studies of head structure

28 Risk assessment 285 Russian research studies

biological effects of pulsed RF radiation 265-286

Saline heating 269 Sampling theorem 48 Scalar function 76 Scalar potential 75, 78, 98 Scalp Laplacian method 73 Scalp potential 81, 85-88, 92, 98-111 Scalp potential field 104 Scalp potential map 87, 102-103, 110-

III Sensory-motor cortex 228, 275-277 Shielding effect 61-62, 66 Shock pulse 157, 165-167, 170-173,

176 Shuttle-box method 280 Signal receptors 150 Signal targeting 57 Signal transduction 182, 196 Signal-to-noise ratio 96 Single layer surface-source model 112 Single-layer model 81 Single pulse 209, 215 Singular value 95 Singular value decomposition (SVD)

97 Singular value decomposition (SVD)

algorithm 88-90, 99, 10 I, III

Sink 79 Skull inhomogeneity 98 Slice selection 44-46 Slice-selective excitation 44, 48, 50 Source 79 Source function 76 Source model 77, 79, 81, 88, 105, 112 Source-conductor model 100 Spatial deconvolution imaging 74, 90,

99 conceptot'88

Spatial summation 213 Spatially extended nonlinear node

(SENN) model 241 Specific absorption (SA) 208, 211 Specific absorption rate (SAR) 13-15,

23,31,33,207-212,215,217,219-221,223,226-228,234-237,240, 247-249,251,267,271-272 peak 25-29 studies on biological materials 24,

31 Specific conductance 122 Specific heat capacity 13-14 Speed oflight I Spermatogenesis 283 Spherical cell model 19, 240 Spin echo (SE) signal 43-44 Spin precession 40, 43, 46, 48, 61 Spin resonance frequency 44 Spin-echo (SE) imaging 48-50, 53, 63 Spin-lattice relaxation 42 Spin-spin relaxation 42 Spin-warp imaging method 44-55 Standardized mortality ratio 243 Standing wave 25-27 Startle reflex 210, 212, 216, 250 Slrophantine K 271 Stun-and-seizure2IO,250 Succinate dehydrogenase 281 Surface charge density 4 Surface compartmental model 150, 188 Surface polarization 18 Surface potential 151-152 Surface-source model 111-1l2 Sympathetic blocker 271

Tachycardia 220-224 Tank model of human head 108

. Taylor expansion 80 ~cm-DTPA (diethylenetriamine

pentaacetic acid) 133-137 Telecommunication I

RF perspectives on 30 TEMPO experimentation 234-237, 245 Temporal constraint 94

301

302

Temporal swnmation 213 Testosterone 281 Tetracaine 272 Tetraethylammonium ion (TEA) 167 Tetrodotoxin (TTX) 167 Thermal damage 170, 182, 197,239 Thermal equilibrium 42 Thermal pain 213, 250 Thermal sensation 213, 250 Thermal stress 230 Thermal tissue damage 148, 169 Three-microelectrode technique 174 Three-shell realistic geometry head

modelllO Three-shell volume conductor model

109 Three-sphere conductor model 81 Three-sphere head model 88, IO I Threshold

absorption 210 auditory 215 channel opening .178-181 electroporation 166, 168-169 membrane potential 169, 177-

178 Thyroid gland 283 Tikhonov regularization scheme 95,

112 Time averaging 207 Time delay 176 Time-of-flight (TOF) method 56 Time-reversing pulse 43 Tissue damage 239 Tracer study 225-226 Transfer function 105, 107 Transfer matrix 88, 90, 92, 103, 105,

108 Transient over-shock 164 Transmembrane current 157,160,163-

165,170-173,177,186,192-193, 195,273

Transmembrane ion flux 196 Transmembrane potential 151-152,

155,158-160 Transmission 24 Transmission coefficient 24 Transporter proteins 183 Truncated SVD pseudo-inverse

method 95 Tumor necrosis 128 Tumor therapy 121-122, 129-130, 143 Turbo gradient spin echo (TGSE)

imaging 54-55 Turbo spin echo (TSE) imaging 57 Twomey regularization method 94

Ultra-wide-band (UWB) radar 208 Ultra-wide-band pulses 209

cardiovascular effects of 248 study of biologiCal effects from 245

Vacuolization 273 Vacuum 3, 12 Vascular spasm 136 Vascular structure 56 Vaseline-gap voltage clamp technique

156,163-164,192,194 Vector of magnetization 40, 43,53 Vectorisation 138 Visual evoked potential 98, 103 Vitality 227, 271 Voltage dependence 162, 194 Voltage sensitivity 153-154,186,197 Voltage sensors 154, 174, 178 Voltage-clamp technique 162 Voltage-gated ion channel 153-154,

157-159, 180 Voltage-sensitive dye 160-161 Voltage-sensitive ion channels 154 Volume conductor 75-76,105

concept of 74-75 model 105

Volume source 74-75 concept of74-75 modeling 76

Walker carcinoma 284 Warmth sensation 213-214 Waveguide exposure system 212, 231-

232 Weighted algorithms 90-92 Weighted general inverse method 90 Weighting factor 91-94, 208 White matter 54 Whole body model 16 Wireless communication 33, 36

advances in electromagnetic fields in living systems

Documents