Nathalie Routhier

Neuro logy and Neurosurgery

McGill University

Montreal, Canada.

September 1997

A thesis submitted to the Faculty of

Graduate S tudies and Research

in partial fblfillment of the requirements of the degree of

Master of Science

O Nathalie Routhier 1997

National Library I * m of Canada Biblioth&que nationale du Canada

Acquisitions and Acquisitions et Bibliographie Services services bibliographiques

395 Wellington Street 395. nie Wellington OttawaON KlAON4 OttawaON KlAON4 Canada Canada

The author has granted a non- L'auteur a accordé me licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loaq distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or elecbonic formats. la forme de microfiche/£ih, de

reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fkom it Ni la thèse ni des extraits substantiels may be printed or otheMrise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

Candidates have the option of including, as part of the thesis, the text of a paper(s)

submitted or to be submitted for publication, or the clearly-duplicated text of a published

paper(s). These texts must be bound as an integral part of the thesis.

If this option is chosen, connecting texts that provide logical bridges between the

different papers are mandatory. The thesis must be Wntten in such a way that it is

more than a mere collection of manuscripts; in other words, results of a senes of papers

must be integrated.

The thesis must still conform to al1 other requirements of the "Guidelines for Thesis

Preparation". The thesis must include: A Table of Contents, an abstract in English and

French, an introduction which clearly states the rationale and objectives of the study, a

cornprehensive review of the literature, a &al conclusion and summary, and a thorough

bib liography or reference lis t.

Additional material m u t be provided where appropnate ( e g in appendices) and in

sufncient details to allow a clear and precise judgement to be made of the importance and

originality of the research reported in the thesis.

In the case of manuscnpts co-authored by the candidate and others, the candidate is

required to make an explicit statement in the thesis as to who contributed to such

work and to what extent. Supervisors m u t attest to the accuracy of such statements at

the doctoral oral defense. Since the task of the examiners is made more difficult in these

cases, it is in the candidate's interest to make perfectly clear the responsibilities of al1 the

authors of the co-authored papers. Under no circumstances can a co-author of any

component of such a thesis serve as an examiner for that thesis.

1 k t want to thank my supervisor, Dr Jean Gotman, who provided me with the

opportunity to conduct this study and to expand my knowledge of the field of

neurosciences. His expertise and his availability were greatly appreciated throughout this

research project.

Dr François Dubeau provided useful comments on various parts of the thesis and

communicated his enthusiasm to the team. 1 want to thank Mç Nicole Drouin, Ms

Lorraine Allard and Ms Josée Ailard for their precious technical support and for their

dedication to the project. Ms Drouin and Ms Josée Allard designed equipment

specifically for this study. M. Eddy Puodziunas greatly helped us in dealing with

technical problems related to the recording of evoked potentials. M. Micheal Frank wrote

a program for the joint display of dipoles and MM. M. Yiou Huang was helpful in

solving computer-related problems. Dr Alejandro Bastos reviewed MRIs and was helpful

in establishing diagnoses. M. Rock Comeau helped us in the difficult issue of mapping

electrode positions in MR1: coordinates.

Dr Daniel Gendron was helpful in getting the experiments started and in organizing al1

testing performed in the EMG department. Dr Felipe Quesney and Dr Terry Peters

provided use of the EEG and MRI facilities respectively.

ABSTRACT

The somatosensory representations of hands and feet were deterrnined using spatio-

temporal source analysis of median and postenor tibial nerve somatosensory evoked

potentials (SEPs). In 10 control subjects, modeled generators mapped on MRI correlated

well with the primary somatosensory cortex identified with slrict radiological criteria For

median nerve stimulation, the main generator was located 3.5 mm fiom the central sulcus

on average ( d e r correction for eccentricity) and was perpendicular to it. For posterior

tibial nerve stimulation, the main generator was slightly antenor to the central sulcus

(average distance 6.6 mm) and pointed predominantly postenorly and towards the

longitudinal fissure. In 3 patients with central dysgenetic lesions, all with normal

somatosensory bctions, modeled generators were located outside the lesion, 2 of these

generators being unequivocally displaced to adjacent normal cortex. The only exception

was one patient with small amplitude SEPs, possibly explained by misaligned pyramidal

cells inside the dysgenetic cortex.

Les représentations sensorielles de la main et du pied ont P.té déterminées à l'aide de

l'analyse spatio-temporelle des générateurs des potentiels somesthésiques évoqués par la

stimulation des nerfs médians et tibial postérieurs. Chez les 10 sujets normaux, les

générateurs illustrés sur 171RM corrélaient bien avec le cortex sensoriel primaire identifié

à l'aide de critères radiologiques stricts. Pour le nerf médian, le générateur principal était

situé i une distance moyenne de 3.5 mm du sulcus central (après avoir comgé

I'excenûicité) et était perpendiculaire à celui-ci. Pour la stimulation du nerf postérieur

tibial, le générateur principal était situé légèrement antérieur au sulcus central (distance

moyenne de 6.6 mm) et pointait principalement vers la fissure longitudinale et l'arrière de

la tête. Chez 3 patients avec lésions dysgénétiques centrales et des fonctions sensorielles

intactes, les générateurs modélisés étaient situés à l'extérieur des lésions, deux d'entre eux

ayant sans équivoque été déplacés vers le cortex normal. La seule exception était un

patient présentant des potentials de très petite amplitude, possiblement causés par le

manque d'alignement des cellules pyramidales à l'intérieur du cortex dysgénétique.

Preface

Acknowledgments

Abstract

Résumé

Table of Contents

1. Introduction 1

1.1 Historkal Review of the Functional Mapping of the Sornatosensory 1

Cortex

1.2. Somatosensory Evoked Potentials (SEPs)

1.2.1 Anatomical Substrate

1.2.2 The Underlying Generators

t -2.2.1 Stimulation of the Median Nerve

2 -2.2.2 Stimulation of the Posterior Tibial Nerve

1 -2.3 Influence of Extemal Variables

1.2.3.1 Variables affecting the Latency of SEPs

1.2.3.2 Variables affecting Scalp Topography

1.3 Investigating the Generators of SEPs: Dipole Modeling

2. Project Definition

3. Literature Review

3.1 Mapping the Somatosensory Cortex

3.1.1 Neurologically Normal Subjects

3.1 -2 Patients with Central Lesions

3.1.3 SEPs in Patients with Central Dysgenetic Lesions

4. Study on the Anatomical Substrate of Generators of Sornatosensory 22

Potentials Evoked by Median and Posterior Tibial Nerve Stimulation

5. Study on Patients with Central Dysgenetic Lesions

6. Conclusion

7. References

1.1 Historical Review of the Functional Mapping of the Somatosensory Cortex

The first large-scale study demonstrating a discrete cortical representation of

somatosensory function in humans was performed by PenfieId and Boldley (1937), using

inmcranial stimulation. From 1928 to 1937, these researchers stimulated the exposed

somatosensory cortex of 126 patients during surgery for the treatrnent of intractable

epilepsy, resulting in either numbness or tingling in parts of the contralateral hemibody-

These sensations were elicited by stimulation of the anterior lip of the post-central gyms,

but stimulation of Brodmann's areas 4 and 6a alpha, anterior to the central sulcus, also

elicited these sensations. A distinct somatotopic arrangement of sensation was invariably

observed on the post-central gyms, with an unequal representation of the diEerent parts of

the body. Penfield and Boldley concluded that cortical representation of cutaneous

sensation is located predominantly but not only on the post-central gyrus and invariably

follows a somatotopic arrangement.

The search for the anatomical substrate of somatosensory function has since been

addressed with other electrophysiological methods. In 1950, Dawson k t performed

scalp-recordings of electrical cortical potentials evoked by penpheral nerve stimulation in

humans. Similar potcntials were later recorded fiom eiectrodes on the cortex, dlowing

for a direct localization of the largest amplitude potentials on cortical anatomy (Woolsey

1958). This paradigrn, called somatosensory evoked potentials (SEPs), approximates

everyday cutaneous stimulation. Penpheral stimulation also evokes magnetic fields

which are generated in the cortical mantle of the sulci, with an orientation perpendicular

t~ the evoked electrical fields. In 1978, Brenner et al recorded for the k s t tirne magnetic

evoked fields (MEF) following cutaneous peripheral stimulation. Uniike SEPs, MEFs are

not distorted by the skull and scalp. However, MEFs only records f?om tangentiai sources

located in the sulci, ignoring the radiai sources. Both SEPs and MEFs results (Yang et

al. 1993; Kawamura et al. 1994; Buchner et al. 1995) were congruent with the homoncula.

organization documented by Pedield. Localization of the somatosensory cortex in area 4

in addition to the post-central gynis has also been reported (Kawarnura et al. 1996).

Penpheral cutaneous stimulation also evokes changes in the metabolic activity of the

somatosensory cortex. Measurements of regional cerebral blood flow or oxygen

consumption with Positron Emission tomography (PET) dunng cutaneous stimulation

show activation of cortical regions involved in somatosensory functions (Fox et al. 1987).

The spatial resolution of this technology now ranges £tom 4.6 to 6.4 mm (Evans et al,

1991), but temporal resolution is quite poor. Functional MRI, which renects variations in

the oxygenation state of the venous vasculature, appears to be a promising non-invasive

technique to study the sornatosensory cortex (Hammeke et al. 1994). The temporal

resolution is approximately 0.5 seconds (Cohen and Bookheimer 1994) and the spatial

resolution of the signal is in the order of 3.75 mm (Haznmeke et al. 1994), and is

expected to reach 1 or 2 mm (Cohen and Bookheimer 1994).

1.2 Somatosensory Evoked Potentials (SEPs)

Amon2 the various techniques outlined above, SEPs present strengths and weaknesses

that detennine its unique contribution to the study of the mapping of somatosensory

cortex. Ln contrast to modern imaging methods, scalp-recorded SEPs constitute a non-

invasive, inexpensive paradigm that provides, with MEG, the greatest temporal resolution

of the above mentioned techniques. Therefore, brain mapping can be detailed in the first

rnilliseconds following stimulus onset. The following paragraphs will detail the technique

of SEPs, its matornicd correlates and the results obtaùied in various studies.

1-2.1 Anatomical Substrate

sornatosensory evoked potentials result kom activation of the dorsal column-media1

Iemniscal system, which is reproduced graphically in figure 1.

Fig . 1 The dorsa l co l umns-media1 lernniscal system

Fig . 2 Brodmann' s areas 1 and 3b on the post-central gyrus

From Mart in , J.H., Jesse l l , T.M. Anatomy o f the Somatic Sensory Systems i n P r i nc ip les o f Neural Science. Edi ted by Kandel , E.R., Schwartz, J.H, Jessell, T.M. New York: E lsev ie r , 1991: 361-364

Electrical stimulation applied at the wrist and ankle to elicit SEPs activates predominantly

large-diameter, fast-conducting group Ia muscle and group II cutaneous afferent fibers

(Halonen et al. 1988). The central process of these fibers ascends in the dorsal columns

ipsilaterally and synapses onto the gracile (legs) and cuneate (arms) nuclei in the lower

medulla. Second-order neurons cross the midline ventral to the central canal as the

interna1 arcuate fibers and form the medial lemniscus, a contralateral ascending tract. The

medial lemniscus synapses onto the ventrobasal complex of the thalamus, more precisely,

the ventral posterior lateral nucleus pars caudalis (VPLC). In the VPL, the legs are

represented more laterally and the arms more medially (Carpenter 1996). Thkd order

neurons consist of thalamo-cortical neurons projecting to Brodmann's areas 1 and 3b on

the post-central gynis (figure 2), where cutaneous sensation is represented (Carpenter

1996).

The cortical representations of the a m and leg are inverted with respect to that of the

thalamus, with the leg and foot represented more medially, in the paracentral lobule, and

the a m and hand more laterally, in the mid-convexity. The spinothalamic system does

not seem to bear much importance in cornparison with the dorsal-column-medial-

lemniscal system in the generation of SEPs (see Chiappa 1989, for review).

1.2.2 The Underlying Generators

1.2.2.1 Stimulation of the Median Nerve

Multiple evoked potentials are generated along the ascending sornatosensory pathway

following stimulation of the median nerve. The potentials labeled N9, NI 3, P 14, N20,

P20, and P22 c m reliably be detected in neurologically normal subjects. The narne o f the

potentials are based on their polarity (in standard recording montages) and their

approximate latency post-stimulus. For example, N9 is a negative potential generated 9

msec after stimulus onset. The nomenclature of the potentials is quite complex since

different authors use different narnes for the same potential- Figure 3 illustrates the early

potentials evoked by stimulation of the nght median nerve. Positivity will be plotted

upwards in the figures demonstrating the various potentials.

r --r - L

FCz

Fig. 3 SEPS following stimulation of the right median nerve in a normal subject. The N13 potential was recorded on the 6" cervical vertebra (C6) and the P l 4 potential could be observed in midline antenor head region (Fz). These potentials were followed by the N20 and P20 potentids, well recorded in the contralateral centroparietal region (CP3) and fkontocentral region (FCl) respectively. The P22 potential was recorded at CP3.

Peripheral and subcortical potentials N9, NT3, and Pl4 will only be discussed briefly as

the emphasis of this study is on cortical mapping. The N9 potential is generated

immediateiy distal to the brachial plexus (Aminoff and Eisen 1996). The NI3 potential,

which is recorded via an electrode piaced on the sixth cervical vertebrae (C6) in reference

to Fz, results either fiom postsynaptic activity in the gray matter of the c e ~ c a l spinal

cord or fkom activity in the cuneate nucleus (Desmedt and Cheron 1980, 198 1). The far-

field P 14 is generated in the medial lemniscus, at the level of the foramen magnum

(Desrnedt and Cheron 198 1; Delestre et al. 1986). It is important to note that an absence

of these potentials does not necessarily impty a lesion in these generators. In fact, factors

such as changes in the resistance or irnpedance of the volume conductor and the anatomic

orientation of the traveling impulse are lmown to influence the generation of these

potentials (Kimura et al. 1983, 1984).

Approximately twenty milliseconds after median nerve stimulation, a positivity (3'20)

develops antenor to the central sulcus and a negativity (N20) is recorded postenor to the

central sulcus in the hemisphere contralateral to stimulation. In figure 3' maximal

amplitude for the N20 and P20 potentials is recorded at electrodes CP3 and FCl,

approximately 20 msec after stimulation of the right median nerve. Two to five

milliseconds later, an isolated positivity called P22 devetops over the left central region.

Nthough some authors locate the generators of the N20, PZ0 and P22 potentials at a

subcorticai level (Chiappa 1989), it is generally agreed that they are cortically generated

potentials (Allison et al. 1989).

Four theories have been proposed to explain the distribution of these scalp-recorded

cortical potentials. In these theories, generators are modeled as dipoles or elecûical

sources that project in opposite directions positive and negative electrical fields of equal

magnitude. First proposed by Brazier (1 949), dipolar rnodek have received some

experirnental support (Landau 1967). Moreover, it is reasonable to assume that

generators are compact neuronal dipo lar sources whenever scalp potential generators are

spatially resû-icted relative to the distance at which they are recorded, regardless of the

underlying electrophysioIogy (Snyder 199 1). Dipoles are usually classified as radial,

tangential or oblique, depending on their orientation with respect to a spherical model of

the head. In the two radial dipoles model, N20 is generated by a radial dipole over the

post-central gyms, while P20 is generated via a separate radial dipole over the pre-central

gyms and other pre-motor areas (Papakostopoulos and Crow 1980). In the tangential

model, a tangential dipole locatsd in area 3b of the post-central gyrus and pointing

towards the pre-central gyrus generates the N20 and P20 potentials (Broughton 1969).

The tangential + somatosensory radial model proposes, in addition to the above discussed

tangential dipole, a radial dipole in the somatosensory cortex (Brodmann's area 1) to

account for the P22 potential (Allison et al. 1980,1989). Finally, the tangential + motor

radial mode1 explains the P22 potential with a radial dipole in the motor cortex rather than

in Brodmann's area I (Deiber et al. 1986; Cohen and Starr 1987; Desmedt et al. 1987;

Nogueira et al. 1989; Abbruzzesse et al. 1990). Some authors hypothesized that N20 arose

fiom a tangentid generator activated by afferents fiom the VPL, nucleus, whereas P22

arose fiom a radial generator in area 4 activated by afferents fiom the nucleus ventraiis

intemedius of the thalamus (Mauguiere et al. 1983; Desmedt et al. 1987; Mauguiere and

Ibanez 1990).

1.2.2.2 Stimulation of the Posterior Tibia1 Nerve

Twenty-one milliseconds after stimulation of the posterior tibial nerve, a negative

potential (N21) is generated at the level of the first lumbar vertebra, probably reflecting

postsynaptic activity produced by axon collaterals in the lurnbar gray matter (Emerson

1988). The N21 potential of posterior tibial nerve SEPs and N9 potential of median nerve

SEPs reflect the integrity of peripheral nerves, allowing for distinctions between

abnormalities of the peripheral and central nervous system. The N35 potential is a small

amplitude negativity broadly distributed over the scalp, and predominating over rnidline

antenor regions (Chiappa 1989) and the contralateral centroparietal cortex (Yamada et

al. 1996). As can be seen in figure 4, a large amplitude positivity (P40) is then maxirnally

recorded at electrode CPz and in an ipsilateral central location (CP3) for stimulation of

the left posterior tibial nerve. A lower amplitude negativity (N40) deveIops over the right

Eonto-central region (FC2, FC4) at nearly the same t h e .

Fig. 4 SEPS following stimulation of the left posterior tibia1 nerve in a normal subject. P40, N50 and P60 potentials were best d e h e d l i the centro-parietal midline region (CPz), while the N40 potential was best defhed in the contra- lateral koonto-central region (FC4).

Three hypotheses have been proposed to explain the generators of N35 and P40. In the

first one, the origin of the generator of N35 is considered to be subcortical. The P40/N40

potentials are generated by a dipole located in region 3b of the contralateral paracentral

lobule. The orientation of this dipole is either tangential or oblique, depending on the foot

and leg representation on the paracentral lobule. The P40/N40 is thought to parallel the

median nenre N20P20 (Seyal et ai. 1983). In the second hypothesis, N35 is also

generated subcortically, but multiple generators underlie the P40/N40 potential (Kakigi

and Shibasaki 1983). Multiple generators are deemed necessary since there is a slight

latency difference between N40 and P40. The third hypothesis postulates that N35, which

is cortically generated, represents the counterpart to the median nerve N20. P40 is

generated by a radial source, rnirnicking the median nerve P22 (Yamada et al. 1996).

However, this theory does not account for the N40 potential which also appears over the

contralateral centroparietal region approximately 40 msec after stimulus onset. Moreover,

if N35 is the counterpart to the N20 generated in the primary sornatosensory cortex, this

potential would be expected to have its maximum amplitude over the cortical

representation of the leg. Yamada et al (1996) reported a maximal amplitude over the

contralateral hemisphere at C3 or C4. Therefore, it is more likely that one tangential or

oblique source or muItiple generators located ia the paracentral lobule generate the

N40P40 potentials, providing a counterpart to the N20P20 rnedian nerve potentiais. No

counterpart has been proposed for the rnedian nerve P22 potential.

1.2.3 Influence of Externa1 Variables

1.2.3.1 Variables affecting Latency

The eEect on SEP latencies of variables such as age, height, gender and antiepileptic

drugs has been extensively studied to improve the use of SEPS as a diagnostic tool.

Results based on absolute latency, which is the time elapsed between stimulus onset and

the potential studied, will not be reviewed here. The emphasis will radier be on central

conduction time (CCT), a synonym for interwave and interpeak latency. CCT is the tirne

elapsed between the latency of a £kt potential generated at the entry or inside the CNS

and that of a potential generated at a later time. CCT is not affected by variables such as

length and temperature of the lirnbs.

Various authon studied the developmental aspect of median nerve CCT (NL3-N20). In

1980, Desmedt and Cheron (l98Oa, 198Ob) observed a lack of correlation between age and

rnedian CCT when comparing 19 octagerÏans to 25 young adults. Two large studies have

since contradicted these results. In 1982, Hume et al (1982) showed, with 83 subjects

aged 10-79, that median CCT remains constant below the age of 50, increases between

the fifth and sixth decades to remain constant thereafter. In a second study, AUison et al

(1983) found a significant increase in the median CCT with age in 286 adults 4 to 95

years old. However, Chu (1986) and Mervaala et al (1988) also conducted large studies

and found that age alone had no significant effect on median CCT. Zegers de Bey1 et al

(1988) replicated these results for the interval P14N20 with a smaller group (N = 40).

For the posterior tibial nenre, Kakigi (1987) found statistically significant shorter

interwave latencies in yomg adults than in elderly subjects. In contrast, Romani et al

(1992) and Kitamura et al (1995) found that age alone did not affect posterior tibial CCT.

Results concerning the effect of age on median and posterior CCT remain thus

controversial.

Although the effect of height on absolute latencies following posterior tibial nerve

stimulation has been well documented, the effect of this variable on interwave latency is

not as clear. Chu (1986) studied SEPS for the posterior tibial nerve and found a

significant correlation between height and tibial CCT. However, two studies contradicted

these results. In the fïrst, Lastirnosa et a1 (1982) observed no significânt height-related

correlation with CCT (N22-P37) in 33 young subjects. In a second study, Kakigi (1987)

found no significant correlation between height and the interpeak latency N28-N35, which

is based on medial lemniscus-sensory cortex CCT. As expected, Mison (1983) found no

significant height-related changes in median interwave latency (N13-N20). These results

were consistently replicated by Hume (1978), Chu (1986), Mervaala et al (1988) and by

Zegers de Bey1 et al (1988), with a slightIy different montage (P14N20). Therefore,

height does not significantly innuence median nerve CCT and the effect of this variable

on posterior tibial nerve CCT is unclear. The difference observed between median and

posterior tibial nerve findings is not surprising since the distance fkom L1 to the

contralateral cenhoparietal cortex is much longer than the distance between C7 and the

cortex-

With regards to the gender variable, Green (1982a) found a signincantly shorter median

CCT (N9-N20) in a group of 21 women compared to a group of 10 men. However, in

two larger series, Alison (1983) and Mervaala et aI (1988) found no significant sex-

related correlations in CCT (N13-N19) in groups of 286 and 120 normal subjects

respectively. These results were replicated by Chu (1986). Since a gender difference was

found with N13-N20 CCT but not with N9-N20, a sex difference in the peripheral CCT

between N9 and N13 potentials might explain this discrepancy.

Hume et al (1979) studied the effect of dnigs on SEPs in comatose patients and found no

correlation between blood levels of phenobarbital(0-146 pg/ml, median 56 pg/ml) and

median nerve CCT. Green et al (1982b) replicated the results obtained for phenobarbital

and found no h g effect with primidone and carbarnazepine in 108 epileptic patients.

However, they observed an ùicreased CCT with a minimal dose of 20 pJml of phenytoin.

Green et al (1982a) have also showed that seizure type, duration of epilepsy, fkequency of

seizures, and abnormal EEG had no effect on central conduction times.

1.2.3.2 Variables affecting Scalp Topography

Few studies have addressed the impact of dernographic variables on the distribution of

SEPs. Kakigi and S hibasaki (1 99 1,1992) investigated the effect of gender and age on

scalp topography of the N20 and P40 component of the SEPs with computenzed bit-

mapped color images of scalp topography. With respect to gender, they reported no

significant difference in scalp distribution of the N20 between female and male subjects,

and significant differences for certain age groups, especially for children and teenagers,

for the PIO. Age represented a more important factor in these studies, although

significant differences between age groups tended to be attributed to differences in

amplitude rather than spatial distribution of the components.

1.3 Lnvestigating the Generators of SEPs: Dipole Modeling

The analysis of surface potentials rernains limited for at least two reasons. First, the

generator is not always located directly under the maximal amplitude potential. Second, a

scalp-recorded potential may result fiom the activiv of multiple generators. Source

analysis evolved out of the concern to characterize the activity of the source itselfrather

than its manifestations.

The characterization of the location and orientation of a source based on its

manifestations on the scalp surface is called the inverse problem. The inverse problem

remains unsolved for vectorial quantities like brain electrical activity. In 1853, Helmholz

stated Siat any potential field on the surface of a volume conductor like the brain is

compatible with an infinite number of interna1 charge arrangements. To address this

limitation, instantaneous and spztio-temporal models yielding a unique mathematical

solution to the inverse problern have been developed.

Instantaneous models use one or multiple dipoles to explain the spatial information in the

potential distribution (in our case SEP) at one instant. In these models, test sources are

placed in a modefed brain and a theoretical distribution of the scalp potentials generated

by these sources is computed. The theoretical potential distribution constitutes the

forward solution to this given set of test sources. The location and orientation of the test

sources are adjusted iteratively w-ith an optimizing algorithm until the forward solution

providing the best mode1 of the actual scalp-recorded potentials is reached. The

corresponding test sources are retained as best models for the generators of scalp recorded

potentials at that instant. Therefore, a unique mathematical solution can be found with

instantaneous models, provided the number of parameters (one dipole has 6 pararneters)

does not exceed the number of recording channels.

Spatio-temporal models have been designed to assess the location, orientation and

ma,+tude of sources generating the observed spatial distribution of scalp potentials

across time. In the moving dipole approach, the dipole which best models activity at each

instant is retained, creating instantaneous models at each instant. The final solution

consists of dipoles, with various locations and orientations, which explain a significant

proportion of the variance in the data at a aven instant. This group of dipoles whose

pararneters Vary across time are often pictured as one dipole migrating through the brain.

This approach is not easily justified in our case, since the generators of SEPS are

presumed to be fixed neuronal masses (Freernan 1975).

Alternatively, spatio-temporal models c m rely on dipoles of static orientation and location

whose strength varies across time. The most commonly used program computing spatio-

temporal models based on fixed sources is named Brain Electrical Source Analysis

(BESA) (Scherg and von Cramon 1985,1986). BESA addresses the inverse problem by

iteratively fitting sources of fixed location and orientation over a given time interval so

that the fonvard solution wili maximally approach the recorded scalp distribution. The

validity of a model is partly judged according to its goodness of fit, which is the

percentage of scalp-recorded variance expIained by the model, as determined by a least-

squares calculation. Thus, a good model is associated with a low residual variance.

To calculate the fonvard solution, BESA models the head as a sphere in which test

charges are passively conducted to the surface of the scalp. Several lines of evidence

support the concept of volume conduction of charges in the brain (Smith et al. 1983), a

concept based on Maxwell's equations. The sphere is actually made of 4 concentric shells

with different conductivities, representing the scalp, skull, cerebrospinal fluid (CSF), and

brain. The radius of the brain is 71 mm, the "concentric layers" of the CSF, skdl and

scalp havins a thickness of 1'7 and 6 mm respectively. As expected, the conductivity of

bone is smallest (0.0042 mho/m), followed by scalp and brain (0.33 mho/m), and CSF

conductivity is highest (1 mho/m). The fissures and ventricles are not represented in this

model, but a simulation study has shown that their effect on the Iocalization of cortical

dipoIes is negligible (He and Musha 1989). Since volume conduction is a linear process,

a linear superposition of the potentials generated by each source is performed to obtain the

forward solution,

In spatio-temporal models, dipole orientation is detennined with more reliability than

location iri the calculation of the inverse solution, since orientation is based on linear

equations and location is computed with non-linear equations (Scherg 1990). Dipole

location c m represent the activity of quite a large region of brain with indeterminate

boundaries and the chosen location represents a 'center of gravity" for activity in this

region (Scherg and Ebersole 1993). Orientation is more accurate and May allow, for

example, the separation of the radial and tangential cortical dipoles of the median nerve

SEPs. The thud characteristic of dipoles, dipole magnitude, partially reflects the

"macroscopic source current in a circumscnbed brain region that contributes significantly

to the surface signals" (Scherg and Ebersole 1993).

Regional sources explain the activity of a circurilscribed brain region with three dipoles

orthogonal to each other. Since the dipoles are orthogonal, any orientation c m be

projected ont0 them and therefore explained by these three dipoles. Regional sources

present the advantage of describing activity in a circumscribed brain region without

having to assess the orientation of the generator. Underlying this approach is the

assumption that the activity modeled is restncted to a circumscnbed anatornical region.

To conclude, both instantaneous and spatio-temporal models correctly avoid necessarily

localizing a source under the maximal amplitude potential, but only spatio-temporal

models rnay disentangle the activity of sources overlapping in t h e , a documented

phenomenon in SEPs (Buchner et al. 1995). Among spatio-temporal models, the moving

dipole approach models sources which diverge fkom the presumed matornical structure of

the neuronal generators. Therefore, only spatio-temporal models relying on static sources

rnay provide adequate answers to the inverse problem applied to SEPs. It must however

be stressed that the most commody used spatio-temporal models are based on the

following assurnptions: The brain is a volume conductor whose shape approximates a

sphere. In the brain, sources are w e l approximated by dipoles with fixed location and

orientation and produce currents which are passively conducted to the surface. The

number of parameters describing the dipolar sources is less than the nunber of EEG

channels.

The k t goal of this study is to assess whether modeled generators of median and

posterior tibial nerve represented in MIU space c m Iocate the primary somatosensory

cortex on the anterior wall of the post-central gyms in neurologically normal subjects.

Strict anatomical cnteria and MRI readings with an experienced radiologist will be used

to define the position of the central sulcus. Mapping of the hand somatosensory cortex

based on the MRI representation of spatio-temporal dipolar models of SEP generators

have achieved reliable localization of the central sulcus in previous studies (Buchner et al.

1 994, 1995). This study will test whether the same spatial accuracy can be obtained with

a slightly different approach based on the use of Current Source Maps to delineate the

epoch to be modeled. This investigation will be extended to the sensory representation of

the foot based on modeling of posterior tibial nerve SEPs. To our knowledge, no study

has addressed this issue with spatio-temporal modeling conjugated with W. The two

analyses Ml1 be performed on the sarne subjects to enable comparisons of different

functional maps in a single subject.

The location, orientation and activity across tirne of modeled sources will be documented

to understand the characteristics of the generators of median and especially posterior tibial

nerve SEPs. For example, basic principles of electroencephalography propose that the

generators of potentials are pyramidal cells oriented perpendicular to the surface of the

brain. If pyramidal cells in a given cortical region can be modeled by a dipolar sheet, it

follows that the modeled sources should be perpendicular to the cortex. To test if models

meet this requirement, the orientation of dipoles modeling the N20P20 potentials of

median nerve SEPs will be to compared to the orientation of the antenor wall of the post-

centrzl gyrus.

Once established in normal controls that source analysis can locate prïmary

somatosensory cortex, the experimental paradigm will be used to locate the

sornatosensory representation of the hand and foot in patients with central dysgenetic

Iesions, who do not always present a clearly identifiable central sulcus.

SEP studies showed that cortical dysgenesis impinging on the central region results in an

abnormal or absent cortical representation of somatosensory fùnction (de Rijk-van Andel

et al, 1992; Vossler et al. 1992; Di Capua et aZ.1993; Raymond et al. 1997). These studies

were based on the analysis of waveforms and did not investigate fùrther the generators of

the potentials, Although source analysis could obviously not be perfomed in the absence

of potentials, a fair number of patients presented abnomally distributed or late potentials

whose generators could have been studied. Progressing towards a more complex analysis,

the present study will investigate the effect of cortical dysgenesis on cortical SEP

generators, using 32 channels EEG recordings, and dipole source analysis combined with

3D-MRI. The patient group will include patients with cortical dysgenetic lesions

irnpinging on the central region. Contra1 subjects will be neurologically normal

volunteers working at the Montreal Neurological Institute. To the best of our knowledge,

such an approach creates a precedent in the investigation of the functional representation

of somatosensory functions in centrally located dysgenetic lesions.

3.1 Mapping the Somatosensory Cortex with SEPS an(

3.1.1 NeuroIogicalIy Normal Subjects

i Source AnaIys

Mapping the central region with dipole source analysis was greatly improved by the

combination of source analysis with skull X-rays and especially 3D-MN. Digitized

electrode positions made possible measurements of eleckode positions in MRI space for

each individual. Individualized electrode positions improved the localization of modeled

sources by canceling the effect of electrode misplacement and determining more realistic

head radii for the spherical head rnodels. Moreover, for the first t h e , the location of

equivalent dipoles could be directly compared to anatomical landmarks

Baumgartner et al (1 993) were the k t to study SEPs with head models based on

electrode positions digitized with respect to skull X-rays. in this coregistration, the sphere

that best fitted the electrode cloud was caIculated and the electrode cloud was then

projected onto that sphere. Source analysis was performed with respect to the electrodes

projected ont0 the sphere and equivalent dipoles were ascertained. The position of

equivalent dipoles was compared only to anatomical landmarks on the s M 1 X-rays so that

function could be superimposed onto an anatornical substrate. However, the anatomico-

functional correlation performed by this group of researchers was limited. The position of

the dipole was compared to anatornical landmarks such as the ear canal and the midline

and not mapped ont0 the skull X-rays. Baumgartner modeled the N20 potential for the

median nerve, the ulnar nerve and the four digits, with a single equivalent tangential

dipole pointhg anteriorly and slightly upwards, explaining 90.4% h of the variance in 4

subjects. Al1 sources were within 15 mm of the C 3 K 4 position of the international 10-20

system and in 2 subjects out of 4, the sequential sensory representation of the digits was

respected.

The combination of source analysis with 3D-MRI for the analysis of median nerve SEPs

was introduced by Buchner et al (1994), in 13 normal subjects and 7 patients with lesions

of the central region (1 angioma and 6 tumors). These authors improved the functiono-

anatomical correlation by mapping the equivalent dipoles directly onto the MRI. They

observed a localization of a regional source less than 3 mm 6om the central sulcus in 15

of 39 conditions (20 subjects x 2 sides of stimulation), and l e s than 6 mm in 20 of 39

conditions (less than 9 mm for al1 subjects). The localization of the regional source was

coniïrmed during neurosurgery in 6 of the 7 patients. These results were reproducible,

since the replications contributing to the average SEP also fell within 9 mm fkom the

central sulcus. Buchner et al's mode1 consisted of two regional sources, one representing

the N14, and the second modeling the N20 and P22. This model explained more than

95% of the variance in the "corticd interval" ie, between the onset of the N20 and the

next positive peak.

Since BESAys inverse solution is based primarily on orientation, the precision of the

localization of equivalent sources has O fien been judged Ïnadequate for preoperative

mapping of the somatosensory region. For example, C u f i et al. (1991) observed an

accuracy of source localization based on scalp potentials in the order of 1 cm at best.

Moreover, Scherg and Ebersole (Scherg and Ebersole 1993) reported that dipoles could be

mislocated by as rnuch as 2.5 cm, yet still image the same temporal waveshapes.

However, the above mentioned study by Buchner et al shows that BESA can provide a

maximal inaccuracy of 9 mm when localizing the equivalent sources with respect to the

centrai sulcus on MRL

To confirm the accuracy of localization of BESA, Buchner et al (1 995) studied the

localization of the five digits on the central sulcus in 8 normal subjects. Their model

consisted of a dipole in the region of the brainstem as weU as two dipoles in the

contralateral centroparietal cortex, one tangential and one radial in orientation. The

cortical sources explained 96.9-98.7% of the variance in the "cortical interval" for the

median nerve stimulation, 90.7-98.2% for the k s t h g e r , 92.3-97.3% for the third h g e r ,

and 92.3-97.7% for the fi& finger. The tangential cortical sources were located within

3 mm fiom the posterior bank of the central sulcus in 14 stimulations, within 6 mm in 10

stimulations and within 9 mm for 8 conditions. The authors reported much greater

variability in the localization of the radial source with respect to the central sulcus.

3.1.2 Patients with Central Lesions

Franssen et al (1 992) used a 3 dipole-mode1 to analyze the rnedian nerve SEPS of 8

patients with small subcortical infarcts located either in the thalamus or in subcortical

white matter underlying the central region (MRI resolution irnpeded a more precise

localization). Five of the 8 patients showed abnormalities in dipole source potentials

associated with cortical potentials on the affected side, compared to the unaffected side.

In three cases, both the tangentid and radial dipoles were diminished in amplitude or

missing. In one case, dipole source analysis clarified the interpretation of the SEPs by

attributing the abnormality to the radial anaor tangentid generator. Upon neurological

examination, the 5 patients with abnormal dipole source potentials presented pure sensory

symptoms or mixed sensory and motor symptoms as opposed to the pure motor symptoms

observed in the 3 patients with normal dipole source potentials. The authors also

examined leWright differences in the location of the dipoles. They found significant

differences only for the radial dipole, in 4 patients. It is impossible to attribute the

abnormal location to matornical or fûnctional differences since the dipoles were not

rnapped ont0 their anatornical substrates.

Correlating function with anatomy, Buchner et al (1994) observed that the localization of

a regional source explaining the cortical activity was less than 9 mm fiom the central

sulcus for 7 patients presenting with cenkal lesions (6 tumors and 1 angioma). The

authors did not report any discrepancy in the localization of the cortical regional source

between patients and nomal controls. Therefore, centrally located tumors of various

histoloag do not seem to drasticdly alter the representation of the somatotopy on the post-

central gyrus.

Dipole source analysis is thus a vahable tool in identi-g abnormalities in the location

of SEP generators. Mapping of dipolar sources onto MRI image has been instrumental in

establishing a solid correlation between fünction and anatomy. Dipole source analysis

also presents the advantage of disentangling abnonnalities due to different generators.

3.2 SEPs in Patients with Central Dysgenetic Lesions

Modem magnetic resonance imaging has largely contnbuted to the increased detection of

cortical dy sgenetic lesions, which O ften underlie epilepsy previously diagnosed as

idiopathic (Kuzniecky et a i 1993; Barkovich et al. 1994). Cerebrd dysgenesis c m be

broadly defined as a cortical malformation resultulg fkom an abnomai neuronal

development Neuronal development is hindered $ different stages of development and

the resulting disorders are classified as pre-migrational, migrational, or post-migrational,

with respect to the onset of neuronal migration. These disorders are associated with

various degrees of cortical andor subcortical abnomalities which presumably deterMine

their effect on SEPS.

In a study on various types of focal cortical dysgenetic lesions, 5/6 patients with abnormal

or absent N20/P20 potentids presented lesions impinging or bordering on primary

somatosensory cortex (Raymond et al. 1997).

In schizencephaly, opened or closed-lip clefts h e d with polyrnicrogyric gray matter

extend from the cortex to the lateral ventricle. In a patient with a closed unilateral parietal

schizencephalic cleft, no N20/P20 were found on the affected side (Vossler et al. 1992).

Hemimegalencephaly is a neuronal developmental disorder characterized by d i f i s e

unilateral hypertrophy of the brain and an enlargement of the gyri, which may present an

abnormal organization. Histologically, thickened gray matter and white-matter gliosis can

be observed, together with giant neurons scattered throughout the cortex and in ectopic

white matter locations. No cortical SEP components were observed on the side of the

dysgenetic cortex in 4 patients with hemirnegalencephaly (Di Capua et al. 1993).

Lissencephaly type 1 is a dif ise developmental disorder characterized by an absence of

cortical circonvolutions. In this four-layered cortex, the gray matter is thickened and a

narrow rim of penventricular gray matter with islands of ectopic neurons is observed. Di

Capua et al (1993) reported an abnormal morphology and prolonged latency for the N20

wave bilaterally in 4 children. However, in a second study, 6 out of 10 children with

lissencephaly type 1 clearly showed an absence of cortical components bilaterally with a

normal N13 (de Rijk-van Andel et ai. 1992).

Astereognosis was the only sensory deficit found upon neurological examination in 3/7

patients with focal dysgenetic lesions (Raymond et al 1997) and 1/ 1 patient with

schizencephaly (Vossler et al. 1 992). Sensory examination was not reported in the studies

on hemimegalencephaly and lissencephaly reported above (de Rijk-van Andel et al. 1992;

Di Capua et al. 1993).

Hernimegalencep haly and lissencephaly are accompanied b y d i f i se abnormal gyral

patterns which most certainly affect the evoked potentials fields. Therefore, both focal

central dysgenetic lesions and di f i se dysgenetic Iesions are associated with abnormal or

absent cortical SEPS, although the effect of diaise cortical dysgenesis in

hemimegalencephaly and lissencephaly cannot be ruled out.

Nathalie Routhier and Jean Gotman

Montreal Neurological Institute & Hospital, and Department of Neurology and

Neurosurgery, McGill University, Montreal, Quebec, Canada.

Address correspondence and reprint requests to Dr Jean Gotmm, Montreal Neurologicd

Institute, 3801 University St., Monireai (QC) H3A 2B4. Tel.: (5 14) 398-1 953 Fax: (5 14)

398-8 106, email: jean@rclvaw.medcor.mcgill.ca

Keywords: Dipole source analysis, somatosensory evoked potentials, media. nerve,

posterior tibia1 nerve, mapping, central region

Spatio-temporal models of the generators of median and postenor tibial nerve

somatosensory evoked potentials (SEPs) were constructed and mapped onto the

individual MIUS. Thirty-two electrode SEP recordings fiom 10 normal individuds were

modeled in a four-shell head mode1 using a regional source approach. For the median

nerve, three-dipole models including one subcortical and two cortical sources explained

93% of the variance in the interval encompassing the N20 and P22 potentials. For the

postenor tibial nerve, two cortical sources accounted for 94% of the variance associated

with the N40R40, NSO and P60 potentials. After correction for eccenhicity, dipole

N2O/P20 was located at an average distance of 3.5 mm (range 0-8 mm) from the anterior

wall of the post-central gynis. The dipole was predominantly tangential, pointing

anteriorly and towards the midline, forming a 75" angle with the post-central g p s .

Dipole N40/P40 was located parasagittally, at an average distance of 6.6 mm and mostly

anterior to the central sulcus. It predominantiy pointed towards the longitudinal fissure

and posteriorly with an orientation ranguig fiom tangential to radial. After correction for

eccentricity, spatio-temporal modeling of SEPs correlated with 3D MRI allows accurate

localization of the central sulcus and provides information on the temporal evolution of

activated somatosensory relay stations.

Spatio-temporal source analysis of somatosensory evoked potentials (SEPs) has

sipnificantly increased the contribution of electrophysiology to the non-invasive rnapping

of somatosensory cortex. Unlike scalp distribution maps (Desmedt and Bourguet 1985)

and instantaneous dipole models (Sutherling et al. 1988) of cortical SEPs, spatio-temporal

modeLing can disentangle the activity of multiple generators overlapping in time, a

documented phenornenon in median nerve SEPs (Baumgartner et al. 1991; Franssen et al.

1992; Buchner et al. 1994, 1995). Moreover, spatio-temporal modeling investigates

solutions in tems of dipolar sources and thus does not necessarily locate generators under

potentials of maximum amplitude. The "paradoxical location" of the P40 potential of

tibial nerve SEPs, which is maximally recorded over the ipsilateral hemisphere but

generated in the hemisphere contralateral to stimulation, demonstrates the usefulness of

this approach.

Spatio-temporal models generally explain early median nerve SEPs with one equivalent

dipote in the region of the brainstem and two dipoles in the contralateral centro-parietal

hand region (Franssen et al. 1992; Buchner et al. 1994, 1995). One of the cortical dipoles

is mainiy tangential, and convera%g evidence corn cortical surface and transcortical

recordings in humans (Allison et al. 1989) and animais (Arezzo et al. 1979) indicates that

it is generated in Brodrnann's area 3b. The second cortical dipole presents a

predominantly radial (Buchner et al. 1995) or inconsisteot (Balmgartner et al. 199 1)

orientation and is thoug,ht to be generated on the anterior crown of the post-central gyrus

(Ailison et al- 1989).

Generators of posterior tibial nen-e SEPs have not been thoroughly investigated with

spatio-temporal analysis. Distribution maps of posterior tibial nerve SEPs based on

surface and intracortical recordings have shown more variability than median nerve SEP

maps, but they mainly suggested a generator located in the paracentral lobule contralateral

ta the side of stimulation, (Lesser et al. 1987; Takahashi et al. 1996). The generator was

thought to be perpendicular to the Iongitudinal fissure, with a horizontal or oblique

orientation, depending on the representation of the foot in the paracentral lobule-

In recent years, mapping of the central region with spatio-temporal modeling was greatly

improved by the combination of source analysis with brain imaging techniques.

Digitizing electrode positions increased the accuracy of electrode positions for each

individual, a crucial step in the detenninatiori of dipole parameters. Baumgartner et al

(1993) modeled the scalp by digitizing the position of the electrodes with respect to skull

X-rays, but no direct correlation between dipole position and its anatomical substrate was

made. They modeled the generator of the N2O potential for the median nerve, the ulnar

nerve and the four digtal nerves. Buchner et al (1994, 1995) improved the fûnctional-

anatomical correlation of SEPs by mapping equivalent dipoles representing the N20

potential of the median nerve and digits directly onto the MRI. For the fïrst time, the

location of equivalent dipoles representing the generators of SEPs codd be directly

compared to the location of primary somatosensory cortex.

Spatial accuracy of spatio-temporal rnodels has largely been criticized in the past since

dipole localization represents the 'tenter of gravity" of an activity generated in a region

which can be quite large. However, using an integrated source analysis-3D MRI

approach, Buchner et a1 (1995) showed that modeled generators for the N20P20

potentials, which were obtained following stimulation of the median and digital nerves,

were located at an average distance of 6 mm and less than 9 mm fiom the anterior wall of

the post-central gynis. Results were repraducible since a maximum distance of 6 mm

separated the location of generators computed in the context of different studies with the

same subjects.

The first goal of this study was to determine the location of the somatosensory

representation of the hand in M N coordinates based on spatio-temporal rnodels of the

generators of median nerve SEPs. The study was designed to alIow replication of

Buchner et al's findings. The use of Curent Source Density (CSD) maps to mode1 the

generators of median nerve SEPs was emphasized. The second goal was to determine the

location of the somatosensory representation of the foot in the brain. To our knowledge,

no study has addressed this topic with spatio-temporal models of posterior tibia1 nerve

SEPs. The two analyses were performed on the same subjects to enable cornparisons of

different fûnctionai rnaps in a single subject. Finally, a cornparison was performed

between the orientation of the modeled generators and the onentation of the pyramidal

cells, which are the most likely generators of the SEPs. This analysis consisted in

assessing whether dipole onentation was perpendicular to the cortical surface in order to

test if models agree with basic principles underlying dipole source analysis.

METHODS AND MATERIALS

Subjects

Subjects consisted of 5 male and 5 fernale nght-handed individuals aged between 25 and

45 years old (mean = 32, SD = 6). None of the subjects presented a history of

neurological disorder or peripheral nerve injury. Informed consent was given by al1

subj ec ts.

M m acqrrisitiorr and digitizafio~z of electrode positions

Global 3D MEUS were acquired using a Philips ACS III 1.5 T haging system (Philips

Medical Systems, The Netherlands). The Tl weighted gradient-echo volume acquisition

(TR = 18ms; TE = 10 ms; a = 30°; RF-spoiled; = 160 slices) with contiguous I mm

sagittal images, was reconstmcted in 3D using a Gyroview graphic workstation.

Electrode positions were digitized using the cornputer guided Viewing Wand system (ISG

Technologies, Mississauga, Canada). Digitization of electrode position allows source

analysis to be based on the actual electrode positions rather than hypothetical positions

derived from the 10-10 electrode system. Anatomical landmarks such as the bridge of the

nose and the left and right traggus valleys were fuçt located with a surgical probe and

were correlated with the same points on the subject's MR image. Electrode positions

were then digitized with respect to these reference points, using the same probe.

SEPs recordi~zgs

Subjects were instructed to lie d o m and relax with their eyes closed. No sedative was

administered and subjects were instructed to stay awake to avoid the effects of sleep on

the amplitude, latency and morphology of early cortical components of SEPs (Nakano et

al. 1995). Left and right median and postenor tibial nerves were electncally stimulated

during 2 sessions of 2000 stimuli each (5 Hz, 0.2 msec). The intensity of the stimulus

was adjusted to obtain a minimal to moderate contraction of the thumb or big toe.

Median nerve responses were recorded using 3 1 electrodes referenced to Fz and placed

according to the 10- 10 system (figure 1).

F i 1 Thirty-one electrode montage used to record rnedian nerve SEPs. For posterior tibial nerve SEPs, Fpl and Fp2 eiectrodes replaced the Fpz electrode, resulting in a thnty-hvo electrode montage. Note the high density of electrodes symmetrically placed over the central region to favor mapping as well as a good fit of the sphencal head mode1 over this region.

For posterior tibial nerve SEPs, Fpl and Fp2 electrodes replaced the Fpz electrode,

resultîng in a thirty-two electrode montage. Elecbodes were placed symmetrically over

b o t . hemispheres, densely covering the central region. Symmetncal electrode coverage is

important since the best fit sphere, upon which the spherical head mode1 is based, should

not be biased by a higher density of electrodes over one hernisphere. A few electrodes

served as peripheral controls: at LI for the posterior tibia1 nerve and on the sixth cervical

vertebra and at Erb's point bilaterally for the median nerve. hpedances were under 5W2.

Recording bandpass was set to 1-300 Hz and responses were amplified by a factor of

10 000 (Meîraco Diagnostics amplifier, Houston). Recording and averaging were

performed with the Impulse software (Stellate Systerns, Montreal). Responses were

sarnpled at a rate of 1024 Hz over an interval ranging f?om 100 msec pre-stimulus to 100

msec post-stimulus. The two replications of 2000 stimuli were averaged provided they

were highly correlated upon visual inspection. Al1 SEPs were also replicated on a 4-

channefs EMGEP recorder in the context of a different study. In the Focus software

(MEGIS, Munich), SEPs were reformatted to an average reference. Intervals ranging

fkom -20 to +40 msec for the hands and £tom -20 to +80 msec for the feet were selected

for analysis in the Besa software (MEGIS, Munich).

Dipole Source Aizalysis

Digitized electrode positions were read in Besa and the sphere best-fitting the electrode

cloud was found. Analyses were performed using four-shell head models, representing

the scalp, skull, cerebrospinal fluid, and brain. Conductivities of the 4 concentric layers

were 0.33 mho/m for the scalp, 0.0042mho/m for the skull, 1 rnho/m for the CSF and

0.33 rnho/m for the brain (Geddes and Baker 1967). Values for the thickness of the layers

were 6 mm for the scalp, 7 mm for the skull, 1 mm for the CSF and approached those

proposed by Stok (1986) for a standard spherical head mode1 (85 mm radius). Values for

the conductivity and thickness of the layers were not adjusted for individual subjects.

SEPS were digitally filtered (30-250 Hz for the hands; 20-150 Hz for the feet) and

baseline corrected, using the mean signal fiom -15 msec to - 5 msec as the baseline.

Channels were discarded if they met the following conditions: 1) the mean signal £tom -

10 to O msec was more than twice the averaged noise in the 32 channels over the same

interval and 2) the signal to noise ratio ( S m ) was less than 1.5. SNRs were calculated by

dividing the root mean square ( R M S ) over the "cortical interval" (see below) by the RMS

£kom -10 to O msec. Using these guidelines, an average of 0.6 electrodes were deleted for

the hand recordings and 1.7 electrodes for the foot recordings, and the maximum number

of electrodes deleted in a single recording was 3.

Dipole modeling was perfomed in intervals d e h e d with global field power (GFP)

(Lehmann and Skrandies 198 O), which is a global rneasure of the amplitude of the

potentials recorded with al1 scalp electrodes at each time sample. Intervals d e h e d by

GFP insure that generaton are modeled over epochs of maximal activity. For the median

nerve, a "subcortical interval" was delineated by minima of GFP before and after the Nt 3

potential. However, GFP did not successfully isolate the N20 and PZ2 potentials fÏom

earlier or later potentials. Therefore, the "cortical interval" was M e r refined by

identiwng, inside the period of maximal GFP, a typical sequence of Current Source

Density (CSD) maps representing the N20, P20, and P22 potentials. CSD maps illustrate

effective sources and sinks of radial current, decreasing the contribution of

circumferential current caused by the low conductivity of the skull (Picton et al. 2935). As

c m be observed in figure 2, the sequence of potentials spanned fÏom the early

manifestations of the N20 and P20 potentials to the beginning of the central P22.

Fig. 2 Sequence of Current Source Density maps illustrating the "cortical interval" chosen for source analysis of the early cortical median nerve SEPS. The cortical interval begins with the apparition of a centroparietal negativity (N20) and a fkonto- centra1 positivity (P20) at 16.1 msec and ends with the apparition of the central positivity f 22 at 20.1 msec. It excludes the farfield NI8 potential seen at 15.2 msec and the N25F25 potentials appearing at 21 msec.

It excluded the Ni8 potential, which has a generator tlïat appears to be iocated below the

thalamus (Maquière and Desmedt 1989). The cortical interval also excluded the N25

and P25 potentials since the location and orientation of the generators of the N20 and P20

were shown to be different f?om those of the N25 and P25 by fitting instantaneous dipoles

at the peak amplitude of the N20/P20 and N25P25 potentials (see results).

The strategy used to mode1 the generator of median nerve SEPS was based on a regional

source approach p u c h e r et al. 1995) and is outlined in figure 3. Regional sources, which

consist of three collocated dipoles orthogonal to each other, present the advantage of

describin~ current flow in a circumscribed brain region in any direction (Scherg and von

Cramon 1986). Therefore, regional sources provide a good starting hypothesis for a more

refined fitting of source location and orientation. The strategy first consisted in fitting a

regional source (ccsubcortical RS") in the early interval to explain the N13 potential. The

first source was maintained in place while a second regional source ("corticzl RS") was

fitted in the cortical interval. The subcortical RS was maintained active because there is

most probably a temporal overlap in the activity of the two generators. Regional sources

were then reoriented to maxirnize the energy explained by the dipoles. Following

reorientation, two of the dipoles of the subcortical RS explained only an average of 5.9%

of the variance in the early interval and one of the dipoles of the cortical RS explained

only an average of 0.4% of the variance in the cortical interval. These dipoles were

considered inactive and were deleted. The orientation and localization of the cortical

dipoles were then simultaneously fitted to r e h e the rnodeling of the cortical generators.

Final models consisted of one subcortical dipole explaining the early interval and two

cortical dipoles, accounting for the activity in the cortical interval. The subcortical dipole

was narned dipole N13 and the two sequentially activated cortical dipoles were narned

dipole N20/P20

31

Posterior Tibia1 Nerve Median Nerve

A) Fitting of SubcorticaIXS

B) Fitting of Cortical RS

C)Reorien tation ofRS

D)Fitting of Cortical Dipoles

Fig. 3 Dipole source andysis strategîes used to model the generators of left median neme and right posterior tibial nerve SEPS. A) Fitting of a regional source in the interval delineated around subcortical potentials. Since subcortical potentids are not easily recorded for the posterior tibial nerve potentials, this was only done for the rnedian nerve potential N13. The k s t source was rnaintained while a second regional source C'cortical RS") was fitted in the cortical interval. B)Fitting of a second regional source in a cortical interval defined around the cortical potentials N20P20 and P22 for the median nerve and around potentials N40P40, N50 and P60 for the posterior tibial nerve. C)Regional sources were reoriented to rnaximize the activity explained by dipoles and inactive dipoles were deleted. D)Fitting of the location and orientation o f the semaining cortical dipoles in the cortical interval to r e h e the modeling of the cortical generators.

and dipole P22, based on the potentials they presumably explained. Dipole N13 aiso

seemed to explain the N14Pl4 potentials, but this simplified name was chosen. A

regional source was used to scan other brain regions for unexplained activity. Ln one

subject, no stable model was found for the two hands and consequently, the data was

excluded fiom analysis.

For the posterior tibial nerve, no "subcortical interval" was defined due to the small

amplitude of the subcortical components. A "cortical interval" was delineated around the

N40, P40, N50 and P60 potentials based on minima of GFP, since a typical sequence of

CSD maps was only observed around 40 ms, and CSD maps were quite variable over the

full cortical interval. Around 40 ms, a parasagittal ipsilateral and midline positivity (P40)

and a contralateral negativity (N40) corresponding either to a tangential or an oblique

dipole were generally seen in the central region. However, an unaccompanied midline

central positivity suggestive of a radial dipole pattern was observed in a few subjects.

Dipole source analysis first involved fitting a regiond source in the cortical interval

(figure 3 j. Dipole orientation was then optimized. One dipole only explained 3.7% of the

variance on average and was therefore deleted, yielding a two-dipole model. One source

reached its first peak activity at the same t h e as the N40P40 potential (dipole N40/P40)

while the second one reached its maximal activity a few milfiseconds later (dipole N50).

As for the median nerve, the orientation and localization of the two cortical dipoles were

simultaneously fitted in the cortical interval to obtain the h a I model. However, the

temporal difference between dipole N4OR 40 and dipole N50 was not maintained when

simultaneously fitting the two sources, since the second source would explain the hi@

energy activity at the beginning of the interval. Unlike dipo le N20/P20 and dipole P22

which had source waveforms of fairly equal energy, dipole N40/P40 explained a large

portion of the variance. To maintaîn the temporal difference between the two sources, a

sub-interval was defined around the second source, at the end of the cortical interval. The

location and orientation of two sources were then simultaneousIy fitted over the full

cortical interval using a 50% variance criterion to maintain the temporal separation of the

two sources. This generated a two-dipole mode1 representing the generatos of the

N40/P40, N50 and P60 potentials,

Late12 cy and amplitude measrîretnents

For the rnedian nerve, the latency of the N13 potenlial was measured at the electrode on

the C6 vertebra. The latency of the first positive peak of the Source Waveform (SWF) of

dipole NI3 was rneasured in the subcortical interval. Latencies of the N20 and P22

potentials were measured at the CP3 or CP4 electrode contralateral to stimulation.

Measurements were taken at the first positive peak of the S WF of dipole N20/P20 and at

the first positive peak of the SWF of dipole P22 in the cortical interval. Latencies of the

P40 and N50 potentiais were measured at the CPz electrode, and if unclear, at the CP3 or

CP4 electrode ipsilaterai to stimulation. Measurernents were taken in the cortical interval

at the first positive peak of the SWF of dipole N40P40 and at the first negative peak of

the SWF of dipole N50. Latencies of smaller amplitude potentials such as the P l 4 and

P20 for the median nerve and the N40 for the posterior tibia1 nerve were not measured.

Amplitude measurements were taken at the peak of the N20 and P40 potentials and

cornpared to baseiine. Interhemispheric amplitude differences were calculated.

Localization of the central sulcus was performed by an experience neuroradiologist on

axial and midsagittal MR images. On axial planes, the superior frontal sulcus, which

separates the supenor and middle frontal gyri, was folIowed postenorly until it fomed a

nght angle with the pre-central sulcus. The central sulcus was identified as the first

sulcus encountered posterior to the pre-central sulcus (Kido et al 1980). On midsagittal

planes, the central sulcus was identified as the notch immediately anterior to the marginal

branch of the cinplate sdcus (Steinmetz et al 1990; Naidich 199 1).

Dipolar sources defined with respect to a spherical head mode1 were superhposed onto

the MRI. For the median nerve, accuracy of Iocalization was detennined by finding the

length of the shortest line between the actual Iocalization of dipoles and the presumed

location of the generators. Distances were evaluated between dipole N20/P20 and the

anterior waIl of the post-central g p s (Brodmann's area 3b) and between dipole P22 and

the crown of the central sulcus (Brodmann' s area 1). The orientation of dipole N20P20

was compûred to the direction of the central sulcus by drawing a line over the segment of

the central sulcus closest to the dipole, and calculating the angle between the line and the

orientation of the dipole. Orientations were compared on a horizontal slice since dipole

N20ff 20 approximates a horizontal dipole and the ventraVdorsal dimension can be

ignored. For the posterior tibial nerve, the positions of dipole N40/P40 and dipole N50

were sirnply compared to the central sulcus, the longitudinal fissure and the surface of the

brain, and the orientations of dipole N40/P40 and dipole N50 were cornpared to the

underlying anatomy. This is because the generators of posterior tibial nerve SEPS have

been less studied and appear to show more variability than the generators of median nerve

SEP S.

Accuracy of the spherical head model

Figure 4 shows the superposition of the spherical head model on the head of an individual

subject.

Fig. 4 Four-shell head model used for source analysis of SEPs. The accuracy of the spherical mode1 was low for temporal electrodes and the neck electrode, which are located far f?om the central region. The spherical mode1 fitted reasonably well the electrodes positioned over the central region.

Temporal electrodes (T3, T4, TS, T6) and Fpz were located 1.5 cm from the surface of the

best fÏtting sphere on average. The neck electrode was 3.6 cm outside the sphere on

average. More importantly, al1 other electrodes were located at an average distance of 3.7

mm f?om the best fitting sphere.

Median N e w e SEPs

Instantaneous rn odeCs for the N20LP20 and N . U ? X potentiais

Instantaneous dipoles were fitted at the peak amplitude of the N20P20 and N2SR 25

potentials to compare the generators of these potentials. Locations of instantaneous

dipoles explaining the N20R20 and N25/P25 potentials were found ta be significantfy

different @ < 0.01) using a 1-tailed, 1 sample t-test based on the distance between the two

dipoles. The average distance bebveen the two instantaneous dipoles was 11 mm (99%

confidence interval ranged fkom 8.1 to 13.9 mm). Lnstantaneous dipoles representing the

N25/P25 potentials tended to be more anterior than dipoles representing the N20P20

potentials, but the difference in the antero-posterior direction did not reach significance on

paired t-tests @ = 0.086). Moreover, the onentation of the two tangentid dipoles was

sometirnes drasticdly different, as can be observed on CSD maps in figure 5.

spacing: 0.5yCI / 0.2SyU/cmrwr2

Fig. 5 Sequence of Current Source Density maps illustrating the difference in orientation between the N20P20 and N25P25 potentials in one subject. A 45O angle separates the two dipoles at 19.1 and 24 msec.

Waveforms and dipole source modeling

Typical SEPs obtained following stimulation of the right median nerve in an individual

subject are illustrated in figure 6. Analysis of median newe SEPs was based on 17/20

recordings, since SEPs were recorded unilaterally in one subject and no valid models

could be produced in another subject.

/

FCz

Fig. 6 SEPS following stimulation of the nght median nerve in a normal subject. The NI3 potential was recorded on the 6h ceMcal vertebra (electrode C6) and the P 14 potential could be observed in rnidline anterior head region (Fz). These potentials were followed by the N20 and PZ0 potentials, well recorded in the contralateral centroparietal region (CP3) and frontocentra1 region (FC1) respectively. The P22 potential was recorded at CP3.

For the median nerve, models based on one subcortical RS and one cortical RS explained

on average 90.8% of the variance in the subcortical interval (SD = 54) and 91.6% of the

variance in the cortical interval (SD = 3.4).

Three-dipole models (figure 3) explained on average 8 1.5% of the variance in the

subcortical intemal (SD = 9), mainly due to dipole N13. Dipole NI3 predominantly

appeared as a low source in the posterior aspect of the head ipsilateral to stimulation. It

pointed anteriorly, upwards and towards the contralateral hemisphere in 12/17 cases.

Three-dipole models aIso accounted for 92.8% of the variance in the cortical interval (SD

= 3 .O). Electrodes for which the residual variance (RV) was fiequently high (RV > 2 x

average RV in a given subject) were 01 or 0 2 ipsilateral to stimulation, the tempord

elech-odes independently of the side of stimulation, and FC3 or FC4 contralateral to

stimulation. Dipole N20/P20 predominantly appeared as a tangential dipole located in the

contralateral centroparietal region (figure 3), but it presented a more oblique orientation in

a few subjects. Fiawe 7 illustrates an example of a three-dipole mode1 in which the

N20P20 dipole was oblique.

Fig. 7 Three-dipole mode1 illustrating the variability in the orientation of the N20F20 dipole, which models the generator of the median nerve N20/P20 potentials. In this subject, the orientation of dipole N20/P20 was rather oblique, unlike the majorîty of dipoles which were tangential.

In 16/27 conditions, dipole N20P20 pointed in an anterior and midline direction. Dipole

P22 reached its maximum activity 2 msec later than dipole N20P20 on average. It was

located in the contralateral parietal or parieto-occipital head regions in 15/17 conditions.

The orientation of dipole PZ2 approached a radial orientation.

Latencies and Amplitudes

Results of statistical cornparisons between peak latencies of source waveforms in 3-dipole

models and SEP cornponents are given in table 1.

* : p < 0.05 Table 1 Cornparison between mean peak latencies of SEPs and mean peak latencies of dipole source waveforms

In 3-dipole models, the peak latency of the source waveform of dipole N13 did not

significantly differ fiom the latency of the N13 potentiai @ > 0.05). The peak latency of

dipole N20P20 was significantly later than the latency of the N20 potential (p < 0.05). A

significant difference was also observed between peak latencies of dipole P22 and of the

P22 potential @ c 0.05). Central conduction tïmes (N13-N20) as well as the lowest peak-

to-baseline amplitude and maximal interside amplitude difference of N2O observed in a

normal subject are given in table 2. The iowest peak-to-baseline amplitude encountered

in control subjects for either median or posterior tibial nerve SEPs was 0.3 1 yV. The

-

- x SD

maximal interside amplitude difference observed in this group was 60% for both median

and posterior tibial nerve SEPs.

, t l

NI3

12.6 1.21

Dip.NP0

19.2 1.18

1 Central conduction time (ms) 1 N20-N13 1 Left 1 5.9 * 0.7

Dip.P22

21.2 1.28

P22

22.3 1.43

1 1 Right 1 5.6 * 0.4

0.00

Dip.Ni3

12.6 1.21

N20

18.4 1.21

P40

39.6

4-95" 1 5.99"

Lowest peak-to-baselïne amplitude

Maximal Interhemisp heric amplitude di fference

Dip P40

39.4

0.70 1.931 2.15

Table 2 Central conduction times, lowest peak-to-baseline amplitude and maximal interside amplitude difference

Arms & legs

Anils

Legs

4.28,

N50

49.4

Dip N50

46.0 2.86

Right N20 &

P40 N20

P40

4.00

15.6 * 1.3 0.31 pV

60%

60%

Anatonticai correlates of modeled generaturs

Regional sources were positioned 5.72 mm fiom the central sulcus on average (SD =

4.55). They were equally distributed on the pre-central and post-central gyri.

Dipole N20P20 was located on average 3.5 mm fiom the anterior w d l of the post-central

gyrus (range = O - 8 mm). Figure 8 shows the close proximity of dipole N20R20 fkom the

anterior wall of the post-central gyms in an individual subject.

Fig. 8 Representation of dipole N20/P20 rnodeling the median nerve N20P30 potentials on the MRI-defined anatomical substrate of an individual subject. The tangentid dipole is located on the antenor bank of the post-central ="ynis, the prirnary somatosensory area- The orientation of dipole N20P20 is towards the midline and the anterior head regions. The position of the hand region along the central sulcus is approximately halfway between the planum temporale and the longitudinal fissure aiong a circumferential path, in a coronal plane.

In 14/17 conditions, dipole N20E20 was located on the post-central gyrus and in the 3

other cases, it was located on the posterior aspect of the pre-central gyrus. The position of

the hand region along the central sulcus was estimated by rneasuring the position of the

dipole along a circumferential path from the planum temporale to the longitudinal fissure

on a coronal plane (figure 8). Dipole N20/P20 was located approximately halfway

(average = 48%, SD = 20%) between the planum temporale and the Iongitudinal fissure.

The depth of dipole N20F20 was variable. In 10 conditions, modeled sources were

located outside the convexity of the hemispheres. This was corrected by projecting the

sources ont0 the surface of the braln. The above described distances fÎom the central

sulcus were cdculated after projection onto the brain surface. Dipole P22 was Iocated on

the post-central gyms in 5/17 cases and in the superior or inferior parietal lobule in the

remaining cases.

In 16/17 subjects, the angle separating the orientation of dipole N20/P20 f?om the

orientation of the closest segment of the central sulcus was 75" on average (SD = 10") and

larger than 55" in al1 cases. Ln the remaining subject, an angle of 6' was observed

between dipole N20P 20 and the central sulcus.

Posterior Tibia1 Nerve SEPs

Wavefornzs and dipole source modeling

Figure 9 illustrates typical SEPs obtained following stimulation of the left posterior tibia1

nerve in an individual.subject.

Fig. 9 SEPS following stimulation of the left posterior tibid nerve in a normal subject. P40, N50 and P60 potentials were best defined in the centra-pariet al midline region (electrode CPz), while the N40 potential was best defked in the contralateral bnto-central region (electrode FC4).

For the posterior tibia1 nerve, models based on one regional source explained 95.3% (SD

= 2.7) of the variance in the cortical interval. Two-dipole models explained on average

93.6% (SD = 3.2) of the variance in the cortical interval. The unexplained RV was most

fkequently attributable to the neck electrode as well as to electrodes Fp2,02, Cz, CPz and

Pz. Dipole N40R40 accounted for approximately 80% of the variance on its own.

Dipole N40/P40 was located in the centro-parietal region contralateral to stimulation, at a

short distance fiorn the longitudinal fissure (figure 3). It consistently pointed towards the

longitudinal fissure and towards the top of the head, with an orientation ranging fiom

tangential to radial. In 17/20 cases, dipole N40P40 also pointed towards the back of the

head at approximately 45 degrees fiom the longitudinal fissure. Dipole N50, which

reached its maximal activity 6.5 miIliseconds later on average, was also located in the

contralateral centro-parietal region in 17/20 cases. Dipole N50 showed a more

inconsistent onentation than dipole N40P40, but it was predominantly horizontal and it

pointed in an anterior direction, approximately parallel to the longitudinal fissure. in 7/20

cases, the orientation of dipole N50 was close to radial.

Latencies and Amplitudes

Dipole N40/P40 became maximally active 39 -4 msec &er stimulus onset on average

(table 1). Its Iatency was not significantly different from that of the P40 potential @ >

0.05). Dipole N50 reached its peak 3.4 msec before the actual N50 potential, a statisticaily

significant Iatency difference (p < 0.05). Table 2 shows central conduction times (N21-

P40) as well as the lowest peak-to-baseline amplitude and maximal interside amplitude

difference obsemed for the P40 potential in a normal subject.

AH atorrr ical currel ares of m odeled gerr erators

The average distance between regional sources and the centrd sulcus was 5.1 mm and

regional sources were located anterior to the central sulcus in 18/20 conditions. Regional

sources were positioned on average ai a depth of 9.5 mm fiom the surface of the brain and

6.1 mm lateral to the longitudinal fissure in the contralateral hemisphere.

Dipole N40R40 was located in the central sulcus or anterior to it (mean distance = 6.6

mm; range 0-18 mm). Figure 10 illuseates the position of dipole N40P40 with respect to

the central sukus in an individual subject.

Fig. 10 Representation of dipole N40/P40 modeling the posterior tibid N40P40 potentials on the MRI-defuied anatomical substrate of a . individual subject. Dipole N40P40 is Iocated anterior to the central sulcus at a short distance £rom the longitudinal fissure. It points towards the longihidinal fissure, the top of the head, and the back of the head, in a direction approaching that of the central sulcus.

Dipole N40/P40 was positioned 2 to 11 mm lateral to the longitudinal fissure (mean = 6.7

mm) and at an average depth of 8-35 mm. Dipole N50 was seated deeper in the brain

(average depth = 15 mm) than dipole N40P40. It was located anterior to the central

sulcus in the majority of cases, and 8.7 mm lateral to the longitudinal fissure on average

(SD = 7).

Gelterators of Median Newe SEPs

After correction for eccentricity, accurate locaiization of primary somatosensory hand

region in the anterior aspect of the post-central gyms can be achieved using spatio-

temporal modeling of the generators of median nerve SEPs. Localization is based on the

position of dipole N20P20, a tangential dipole modeling the early cortical components of

median nerve SEPS. The average distance between dipole N20/P20 and the anterior wall

of the post-central gyrus was 3.5 mm and the maximal distance was 8 mm. These resdts

are comparable with those of Buchner et al (1995) who reported an average distance of 6

mm and a maximal distance of 9 mm between modeled generators for the N20/P20

potentiais and Brodmann's area 3b. In addition, the position of the hand region dong the

central sulcus was found to be approximately halftvay between the planum temporale and

the longitudinal fissure dong a circumferential path, in a coronal plane. This last measure

constitutes a gross estimation of the localization of the hand region dong the post-central

g y m since it is based on a coronal plane which ignores the circumvolutions of the post-

central gyms. However, it is interesting to note that in all cases, dipole N20P20 was

located in a f d y narrow region between 27% and 65% of the distance fiom the planum

temporale to the longitudinal fissure.

In contrast, dipole P22 constitutes a poor marker of the somatosensory hand region.

Dipole P22 was located on the post-central gynis in only 29% of the cases and in the

idenor and supenor parietal lobules in 71% of the cases. One possible explanation for

this unexpected finding is the length of the cortical interval which includes only part of

the activity of the source wavefom of dipole P22. This short interval may lead to an

inaccurate localization of dipole P22. However, the source was also frequently located in

posterior head regions when, in a preliminary analysis, we used a longer cortical interval

including the N20P20 and N25/P25 potentials (Routhier and Gotman 1996). A longer

interval was not associated with a localization of dipole P22 on the crown of the post-

central gyrus, as predicted by Allison et al (1989). Therefore, the present results do not

provide strong support for the hypothesis of a second cortical generator located on the

post-central gyrus. Buchner et al (1 995) reported an equal distribution of the radial dipole

antenor, posterior, or within the central sulcus, but these authors did not report the actual

distance fiom the crown of the post-central gyms.

Three-dipole models based on one subcortical dipole and two cortical dipoles explain well

the scalp-recorded potentials evoked by stimuIation of the median nerve. Eighty percent

of the variance in the subcorticul interval was explained by a low source in the posterior

aspect of the head. Dipole NI3 models the large amplitude N13 potential, since peak

latencies of the source waveform and of the potential do not significantly differ. It also

appears to mode1 the far-field P 14 because the dipole points towards anterior head regions

where P l 4 is maximally recorded, The position of dipole N13 is not weli defined due to

the scarcity of eiectrodes in the inferior haif of the sphere. Moreover, the genesis of this

far-field potential is likely linked to a variety of factors: fïxed neural generators, changes

in the resistance or impedance of the volume conductor, and the anatomic orientation of

the traveling impulse (Kimura et al. 1983 1984).

Early cortical components of median nerve SEPS are adequately modeled by dipole

N20P20 and dipole P22, as shown by the very large portion of the variance explained by

three-dipole models in the cortical interval (92.8%). However, the peak activity of the

N20P20 and P22 dipole source waveforms differ significantly fiom the peak activity of

the N20/P20 and P22 potentials. Dipole N20P20 peaks slightly but quite consistentty

Mer than the N20 potential, measured at the CP3 or CP4 electrode contralateral to

stimulation. This fhding was also reported by Franssen et a1 (1992) and Buchner et al

(1995). Measurements at the CP3 or CP4 electrode represent the largest amplitude

potential, but this high amplitude activity is restricted to a few electrodes. Approximately

one millisecond later, the activity fiom the tangential generator is spread to neighboring

electrodes, resulting in a higher global field power. Since source analysis explains mainly

latencies of maximum global field power, the activity of dipole N20P20 peaks

approximately 1 msec after the CP3/CP4 maximum. With regards to dipole P22, the

maximum activity occurs before the P22 potential measured at the CP3 or CP4 electrode

contralateral to the side of stimulation. When a CSD map is constructed at the peak

latency of the P22 potential (measured at the CP electrode), we fiequently observe a

central positivity and an 'inverted tangential dipole' corresponding to the N25P25

potentials (figure 2: CSD map at 21 msec). Since the N25/P25 potentials were excluded

fkom analysis, dipole P22 will obviously not mode1 the activity at the peak latency of the

P22 potential. In our opinion, dipole PZ2 models the transition between the N20P20 and

N25/P25 potentials which appears as a central positivity often accompanied by a

negativity posterior to the N20 potential (figure 2: CSD map at 20.1 msec). This

explmation would account for the posterior localization of dipole P22. The transition

state would also explain the variability that we and others have observed in the orientation

of dipole P22 Paumgartner et al. 1991).

Three-dipole models explained nearly dl data sets (17/19), and the two cases in which

these models were not applicable occurred in a single subject who presented one of the

worst overall signal-to-noise ratio.

CSD maps have proven usekl in the definition of the cortical interval of median nerve

SEPS by allowing the exclusion of undesired potentials. The analysis of the position of

instantaneous dipoles fitted at the peak amplitude of the N20P20 and N25P25 potentials

showed a significant difference between the position of the two dipoles @ < 0.01).

However, this analysis is based on instantaneous dipoles which may lead to erroneous

results since the activity of multiple sources overlap in time in the cortical interval.

Therefore, a different approach based on spatio-temporal modeling was taken to confïnn

our findings. ThÏs approach consisted in testing how well three-dipole models based on

the cortical interval explain the interval covering the N25R25 potentials. In 9/17 cases,

the fixed dipoles explained well the N25P25 potentials, resulting in a less than 5%

increase in RV cornpared to the cortical interval. However, in 6 cases, the RV increased

by at least IO%, reaching a 28% increase in the case illustrated in figure 5. The

inadequacy of the modeled generators of the N20 and P22 potentials to explain the later

potentials was obvious in 6/17 cases. Therefore, the cortical in t end should not include

the N25@25 potentials which appear to be generated by a generator different from the

N20/P20 potentials. This fhding disagrees with the positive-negative potential sequence

documented in animal experiments (Mitzdorf, 1985). In addition, there was a tendency for

instantaneous dipoles representing the N25P25 potentials to be located more anteriorly

than the generators of the N20/P20 potentials. This was also observed by Garcia-Larrea et

aL(1992) who reported a more anterior isoelectric line for dipole P24/N24 than for dipole

P20N20. This suggests that the generator of the N25/P25 potentials may be in the

posterior wali of the pre-central gyrus, which is known to have strong anatomical

connections with the post-central gynis.

Generators of Posterior Tibiul Nerve SEPs

Spatio-temporal modeling of the generators of posterior tibial nerve SEPs localize the

modeled generators near the primary somatosensory foot region. Dipole N40/P40 was

located at an average distance of 6 mm fkom the central sulcus or its projection on the

paracentral lobule. A position 2 to 11 mm lateral to the longitudinal fissure and an

average depth of 8.4 mm were observed. This location is shallower and closer to the

rnidluie than the position of equivalent current dipoles found with MEG at this latency

(Huttunen, 1987). Dipole N50 was seated deeper in the brain and slightiy more lateral

than dipole N40P40. A deep seated dipole was suggested by Takahashi et a1 (1996) to

explain two different wave patterns at superficial and deep locations along the

longitudinal fissure. Unexpectedly, dipole N40ff40 and dipole N50 were located anterior

to the central sulcus in the rnajority of cases. The reason for this anterior location is

unclear. It is unlikely due to an inadequate head model since sources for the hanci SEPs

based on the same head model did not display such a bias.

The N40P40, N50 and P60 potentials of posterior tibial nerve SEPs are well accounted

for by two-dipole rnodels, since 94% of the variance in the cortical interval is explained

on average. Al1 recorded data sets could be explained by simple two-dipole models. At

the beginning of the interval, dipole N40R40 is responsible for the generation of the

N40P40 dipole since dipole N50 becomes ody active 6.5 msec later. These results are in

agreement with models generated fiom MEG recordings (Hu~inen et ul. 1987) and

topographical analysis of CSD maps (Nagarnine et al. 1992) following stimulation of the

posterior tibial nerve, which point to a single generator at these latencies. Even though

dipole N40/P30 accounts for 80% of the variance in the cortical interval, a second source,

dipole MO, is needed to explain potentials observed later than the N40/P40 potential.

This again is in agreement with MEG models which cannot explain later potentials with a

single equivalent dipole (Huttunen et al. 1987).

Unlike modeled generators for the early corncal median nerve potentials, the activity of

dipote N40P40 peaks at the same t h e as the P40 potential. Moreover, the spread of

activation is restncted and no subsequent increase in global field power is seen. These

three observations support the hypothesis that dipole N40P40 fully explains the P40

potential. Dipole N50 reaches its first peak activity before the N50 potential and models

the N50 and P60 potentials.

Even though good agreement is observed between recorded potentials and forward

solutions based on modeled generators, these models constitute simple approximations of

the generators of SEPS. The remaining residual variance at certain electrodes testiQ to the

complexity of the neuronal generators of these potentials.

Accwacy of Head models

Ln the present study, dipole N20P20 occupied a very eccentric position in some subjects.

This modeling error was not observed for dipole N40P40, but dipole N40F40 was still

located more superficially than the corresponding MEG dipole, which was located at a

depth of 27 & 4mm (Huttunen, 1987). The larger eccentricity of dipole N20/P20

compared to dipole N40/P40 is probably due to the increaçed inaccuracy of the head

mode1 fiom media1 to lateral aspects of the head (figure 4). This inaccuracy was

compensated by projecting dipole N20P20 ont0 the brain surface. However, an overall

tendency of dipoles to occupy eccentric positions was also observed. The inaccuracy in

the depth of the dipoles was not observed when prelimùiary results were computed with

the Ary approximation to the three-shell head mode1 (Routhier and Gotman 1996). Four-

shell head models were chosen since various authors (Berg and Scherg 1994; Zhang and

Jewett 1993) argued that the Ary approximation could lead to inaccurate location for

sources located at large eccentricity. By adopting Cshell head models, the eccentricity of

modeled generators increased, but the distance between the 'projected dipole N20/P207

and the central sulcus dong the antero-posterior and medial-lateral axes was reduced,

Orientation of the gelterators

Regional sources also provide good approximations of the location of the plimary

somatosensory areas. Regional sources located the cenbal sulcus with an average

accuracy of 5.7 mm for the median nerve SEPs and 5.1 mm for the posterior tibial nerve

SEPs.

However, regional sources lack important information that c m contribute to the

characterization of the generators: the orientation of the dipoles. The orientation of the

rnodeled generators can be compared to the underlying anatomy to assess whether

modeled generators are onented perpendicular to the cortical surface, as wouid be the case

if dipoles modeled sheets of pyramidal cells. For the median nerve, dipole N20P20

consistently pointed towards the midline and anterior head regions and it approximated a

horizontal dipole. An average angle of 75' separated the orientation of dipole N20/P20

from the orientation of the closest segment of the anterior wall of the post-central gyms.

We can thus deduce that the orientation of dipole N20P20 differs Erom the presumed

orientation of pyramidal cells by an average angle of 15". Even though the rnodeled

generator is not exactly perpendicular to the anterior wall of the post-central gyrus, it

constitutes a good approximation of the orientation of the pyramidal sheets in the primary

somatosensory area. Such an analysis was not performed for dipole P22 which presented

an inconsistent orientation.

For the posterior tibial nerve, dipole N40P40 consistently pointed postenorly, medially

and towards the top of the head, with an orientation ranging fiom tangential to radial.

Cornparisons between dipole orientation and the orientation of the anatomical substrate

are difficult because dipole N40/P40 is f?equently oblique and it cannot be well visualized

on either one of the t.hree planes. In cases in which dipole N40P40 is tangential, the

orientation closely resembles the orientation of the central sulcus.

Localization would be improved by constraining dipolar sources to be perpendicular to

the cortical surface. For example, in one subject, dipole N20P20 was found to be almost

parallel to the direction of the central sulcus (figure 11).

Fig. 11 Cornparison between the orientation of the tangential dipole N20P20 and the orientation of the central sulcus in an axial slice. In this unusual case, the orientation of dipole N20/P20 is almost parallel to the orientation of the central sulcus measured at the segment closest to the dipolar source. This hd ing disagrees with basic ideas underIying source analysis whïch suggest that dipole orientation represents the orientation of pyramidal cells, perpendicular to the surface of the cortex. However, a slight change in the location of the dipole towards a more anterolateral position would result in an orientation nearly perpendicular to the orientation of the central sulcus.

However, a segment of the anterior wall of the post-central gyms located approximately 5

mm lateral to the modeled source, is quasi perpendicular to the orientation of the dipole.

Therefore, source analysis based on a normality constraint would have found a source

slightly more laterally, in good agreement with the anatomical substrate.

In conclusion, spatio-temporal rnodeling of median nerve and posterior tibia1 nerve SEPS

based on a relatively small number of electrodes provides an effective tool for the

localization of the central sulcus. The digitization of electrode coordinates in MRI space

allows for the correlation of source localization with the MRI-defined anatomical

substrate. Constraining generators to be located in the gray matter with an orientation

normal to the cortex will M e r Uicrease the mapping power of this technique, making it

applicable to a wide variety of evoked responses. Although adequate mapping of

somatosensory cortex can be achieved with spatio-temporal rnodeling, extemal validation

with high spatial resolution mapping techniques in the sarne MR space is warranted. In

the context of this rnultimodality mapping, spatio-temporal modehg will be instrumental

in interpreting the temporal evolution of the sensory systems activated.

Acknowledgrnents: We are gratefûl to Ms Nicole Drouin and to Ms Lorraine Allard for

technical assistance and for their dedication to this research project. We also want to

thank Dr François Dubeau for useful comments on the manuscript and for constant

support. Dr Daniel Gendron, Dr Terry Peters, and Dr Felipe Quesney provided use of the

EMG, MN, and EEG facilities respectively. Dr Alejandro Bastos helped in the

interpretation of MRls-

l This project was supported by the Savoy Foundation and the Neurology and

Neurosurgery program at McGill University.

Allison, T., McCarthy, G., Wood, CC., Darcey, T.M., Spencer, D.D., Williamson, P.D.

Human Cortical Potentials Evoked by Stimulation of the Median Nerve. 1.

C ytoarchitectonic Areas Generating S hort-Latency Activity. J Neurophysiol, 1 98 9, 62:

694-710.

Arezzo, J., Legatt, A.D., Vaughan, J., H.G. Topography and intracranial sources of

somatosensory evo ked po tentials in the monkey. 1. Early components. Electroencep h clin

Neurophysiol, 1979,46: 155-1 72.

Baumgartner, C., Barth, D.S ., Levesque, M.F., Sutherling, W. W. Functional anatomy of

human hand sensonmotor cortex kom spatiotempord analysis of electrocorticography.

Electroenceph clin Neurophysiol, 199 1, 78: 56-65.

Baumgartner, C., Doppelbauer, A., Sutherling, W.W., Lindinger, G., Levesque, M.F.,

Aull, S., Zeitlhofer, I., Deecke, L. Somatotopy of Human Hand Somatosensory Cortex as

Studied in Scalp EEG. Electroenceph clin Neurophysiol, 1993, 88: 27 2 -279.

Berg, P., Scherg, M. A fast method for foward computation of multiple-shell spherical

head rnodels. Electroenceph clin Neurophysiol, 1994,90: 58-64.

Broughton, R.J. Somatosensory Evoked Potentials in Man: Cortical and Scalp

Recorduigs. Unpublished doctoral dissertation, McGill University, Montreal, Canada,

1967.

Buchner, H., Adams, L., Knepper, A., Rüger, R., Laborde, G., Gilsbach, LM., Ludwig, I.,

Red, I., Scherg, M. Preoperative localization of the central sulcus by dipole source

analysis of early somatosensory evoked potentials and three-dimensional magnetic

resonance imaging. J Neurosurg, 1994,SO: 849-856.

Buchner, H., Adams, L., Müller, A., Ludwig, L, Knepper, A-, Thron, A., Niemaun, K.,

Scherg, M. Somatotopy of Huma. Hand Somatosensory Codex Revealed by Dipole

Source Analysis of Early Somatosensory Evoked Potentids and 3D-NMR Tomography.

Electroenceph clin Neurophysiol, 1995, 96: 12 1-134.

Desrnedt, J.E., Bouguet, M. Color imaging of parietal and frontal somatosensory

potential fields evoked by stimulation of rnedian or postenor tibial nerve in man.

Electroenceph clin Neurophysiol, l985,62: 1-17.

Franssen, H., Stegeman, D.F., Maleman, J., Schoobaar, R.P. Dipole Modelling of Median

Nerve SEPS in Nomal Subjects and Patients with Small Subcortical Tnfarcts.

Electroenceph clin Neurophysiol, 1992, 84: 461-417.

Garcia-Larrea, L., Bastuji, H., Mauguiere, F. Unmasking of cortical SEP components by

changes in stimulus rate: a topographic study. Electroenceph clin Neurophysio l, 1992,

84: 71-83.

Geddes, LA., Baker, L.E. The specific resistance of biological materials - a compendium

of data for the biomedical engineer and physiologist. Med Bi01 Eng, 1967,s: 271-293.

HuMuien, J., Kaukoranta, E., Hari, R Cerebral magnetic responses to stimulation of tibial

and sural nerves. J Neurol Sci, 1987,79: 43-54.

Kido, D.K., Le May, M., Levinson, A.W., Benson, W.E. Computed tornographic

localization of the precentral gyms. Radiology, 1980, 135: 373-377.

Kimura, J., Mitsudome, A., Beck, D.O., Yamada, T., Dickins, Q.S. Field Disiribution of

Antidromically Activated Digital Nerve Potentials: Model for Far-Field Recording.

Neurology, 1983,33: 1164-1169.

Kimura, J., Mitsudome, A., Yamada, T., Dickins, Q.S. Stationary Peaks koom a Moving

Source in Far-Field Recording. Electroenceph clin Neurophysiol, 1984,58: 351-361.

Lehmann, D., Skrandies, W. Reference-kee identification of components of

checkerboard-evoked multichanneI potential fields. Electroenceph clin Neurop hysio 1,

1980,48: 609-62 1,

Lesser, R.P., Lüders, H., Dinner, D-S., Hahn, J., Morris, H., Wyliie, E., Resor, S. The

source of 'paradoxical lateralization' of cortical evoked potentials to posterior tibial nerve

stimulation. Neurology, 1987, 37: 82-88.

Mauguiere, F ., Desrnedt, J.E. Bilateral Somatosensory Evoked PotentiaIs in Four Patients

with Long-standing Surgical Hemispherectomy. Ann Neurol, 1989,26,6: 724-73 1.

Mitzdorf, U. Current source-density method and application in cat cerebral cortex:

investigation of evoked potentials and EEG phenomena. Physiol Rev, 1985, 65: 3 7-100.

Nagamine, T., Kaji, R., Suwazono, S., Hamano, T., Shibasaki, H., Kimura, J. Cunent

source density mapping of somatosensory evoked responses following median and tibia1

nerve stimulation. Electroenceph clin Neurophysiol, 1992, 84: 248-256.

Naidich, T.P. MR imaging of brain surface anatomy. Neuroradiology, 1991, 33(Suppl):

S95-S99.

Nakano, S., Tsuji, S., Matsunaga, K., Murai, Y. Effect of sIeep stage on somatosensory

evoked potentials by median nerve stimulation. Electroenceph clin Neurophysiol, 1995,

96: 385-389.

Picton, T.W., Lins, O.G., Scherg, M. The recording and analysis of event-related

potentials. In: F. Boller, J. Grafman (Eds), Handbook of Neuropsychhology, Ch. 1. Elsevier

Science B.V., Amsterdam, 1995: 3-73.

Routhier, N., Gotrnan, J., Anatomical Substrate of Generators of Somatosensory

Potentials Evoked by Posterior Tibiai Nerve Stimulation. American Clinical

Neurophysiology Society Annual Meeting, 1996, (Abstract), (in press).

Scherg, M., von Cramon, D. Evoked Dipole Source Potentials of the Human Auditory

Cortex. Electroenceph clin Neurophysiol, l986,65: 344-360.

Steinrnetz, H., Furst, G., Freund, H.J. Variation of perisylvian and calcarine anatomic

landmarks withh stereotaxic proportional coordinates. AJNR: Am J Neuroradio, 1990,

11: 1123-1 130

Stok C.J. The inverse problem in EEG and MEG with application to visual evoked

responses. CIP Gegevens Koninklijke Bïbliotheek, The Hague, 1986.

Sutherling, W.W., Crandall, P.H., Darcey, T.M., Becker, D.P., Levesque, M.F., Barth,

D.S- The magnetic and electric fields agree with intracranial localizations of

somatosensory cortex. Neuology, 1988,38: 1705-1714.

Takahashi, H., Suzuki, I., Ishijima, B. Cortical and Subcortical SEPS Following Posterior

Tibia1 Nerve Stimulation. Brain Topography, 1996, 8, 3: 233-235.

Woolsey, C.N. Organization of Somatic Sensory and Motor Areas of the Cerebrat Cortex.

In: H.F. Harlow, C.N. Woolsey (Ed.) Bio&ical and Biochenzical Bases of Behavior,

University of Wilconsin Press, Wisconsin, 1958: 63-8 1.

Zhang, Z., Jewett, D.L. Insidious errors in dipole localization parameters at a single time

point due to mode1 rnisspecification of number of shells. Electroenceph clin

NeurophysioI, 1993, 88: 1-1 1.

METHODS

Subjects

Subjects consisted of 1 femde (case 1) and 2 males (cases 2,3) aged 20,30, and 35

respectively. Patients were s u f f e ~ g fiom medically intractable epilepsy and were

undergohg an investigation to determine whether they were surgical candidates. A

description of dernographic data, seinire type, MRI hdings, and neurological symptoms

is given in table 1. The three patients presented no sensory deficits upon neurological

investigation. Patients were taking antiepileptic drugs at the time of the examination.

One of the patients (3) was taking Dilantin, but no increase in central conduction time

(CCT) was observed, contrarÏly to reports by Green et al (1982b). Factors such as seizure

type, duration of epilepsy, fkequency of seimes, and abnormal EEG were not controlled

for because no effect on central conduction times has been reported (Green et al. I982a).

Table 1 Demog PNH = periveni dysplasia; C = c

1

2

Sex

F

M

Handed ness R

Motor & sensory deficits

Age

20

30 R - .

rapid altemating & fine finger mov. of

lefi hand

%TRI-defined Lesion

lefi anterior pre- cuneus FCD

S eizures

PC with contraction of ri& arm and deviation

nght pensylvian PMG + bil. PNH

generalized tonic clonic

of head to right PC with feeling of straageness + GTC

R

aphic data, seizure type, MRI findings, and neurological symptoms icular nodular heterotopia; PMG = polyrnicrogyria; FCD = focal cortical mtral; P = parietal; O = occipital; PC = partial complex; GTC =

lefi C-P + right P-O PMG + infolding in

lefi C region + abnormal gyral pattern right C

tingling in right hemi- body+convu!sions in

right hernibody

nght spastic herniparesis and

hemiatrophy

Radiologicnl classification of dysgenetic lesions and M X l rneasurements

Al1 classincations were made on radiological criteria Some patients presented more than

one type of dysgenetic lesion (table 1). Pen~en~cuZar nodular heterotopia (PNTI),

characterized by nodules of gray matter isointense to the cortex and lùiiog the lateral

ventricles, was observed in patient 2. Multiple small irregular gyri, irregularities of the

-white matter interface, and normal cortical thiclaiess characterized polynzicr~gy~a

(PMG) (patients 2 and 3). In patient 3, a severe infolding lined with PMG was observed

in the left central region. Focal cortical dysplasia (FCD) seen in patient 1 was

characterized by an abnomal cortical pattern, cortical thickening and blurrïng of the gray

mattedwhite matter interface.

In patients 1 and 2, the central suIcus could be identified ushg the same critena as was

used in control subjects. In patient 3, it was ill-defined in the right hemisphere and

unidentifiable in the le fi hemisp here

It was determined whether lesions invaded the primary somztosensory representations of

the hand and foot on the post-central gyrus. Based on control subjects, the extent of the

hand rspresentation on the post-central gyms was grossly d e h e d as 48% t 20% of the

distance fiom the planum temporale to the longitudinal fissure on a coronal slice taken at

the level of the post-central a"ynis. The foot representation occupied a parasagittal

position close to the paracentral lobule and less than 1.7 cm away fiorn the longitudinal

fissure in 95% of the control cases. In patient 3, the bilateral lesions invaded the central

region in a d i f i se marner, and even though the central sulcus could not be identified, al1

parts of the "ill-defined post-central gyri" were considered to be invaded by dysgenetic

lesions (or by abnormal gyral patterns).

SEPs

Recording parameters, processing of SEPs, dipo le modehg, and measurements were the

same as in the study on control subject unless specified otherwise in the following

paragraphs.

Laten cy and Amplitude Measurernents

In cases of abnormally distributed potentids, latencies and amplitudes were measured at

the highest amplitude peaks. Patients central conduction times (CCTs) which were more

than 3 SD from the mean CCT of the control group were considered abnorrnal. Analysis

extended to 40 ms post-stimulus for the median nerve and 80 ms for the posterior tibia1

nerve SEPs. Interhernispheric amplitude differences exceeding the highest difference

fomd in control subjects were considered abnormal. Peak to baseline amplitudes lower

than the lowest amplitude value encountered in control subjects were classified as

abnormal. However, it is more difficult to interpret peak to baseline amplitudes given the

hi& variability observed in the control group.

Distribution of SEPs

The distribution of early cortical SEPs was analyzed with Current Source Density (CSD)

maps based on the 32 electrode recordings and compared to typical maps of control

subjects.

SEPs

Normal SEPs were recorded at penpheral locations (C6 and LI), confimiing the absence

of abnormalities of the peripheral nervous system. As can be seen in table 2, SEPs were

normal in patients 1 and 2. Patient 3 presented srnaIl and ill-dehed N2OP20 and

N40/P40 potentids for the right median and posterior tibial nerve SEPs. For the left

median and postenor tibial nerve SEPs, early cortical potentials were of normal amplitude

and latency, but CSD maps showed a parietal rather than a centrai location.

SEPs vs Location of Dysgenetic Lesians

Table 2 shows a cornparison between the part of the post-central gyms invaded by the

dysgenetic lesion and SEP abnormalities.

MRI Part of post-central gyrus (PCG)

SEP abnormalities L. median 1 R median 1 E. 1 R

1 2 3

Table 2 CSD = CSD maps sequence; L = latency; A = interside amplitude diffience, a =

invaded by dysgenetic Iesion -

right PCG ( with large infolding on left PCG)

absolute amplitude

hand region on right PCG hand and foot regions on left and

No SEP abnormalities were observed in patient 1, who presented focal cortical dysplasia

bordering, but slightly outside primary somatosensory cortex. The polymicrogyric Lesion

of patient 3 clearly invaded the hand and foot regions of the lefi post-central gyms and

was accompanied by a large infolding. In the right central region, an abnorrnal gyral

pattern, which was not clearly dysgenetic in nature but which may be an extension of the

parieto-occipital PMG, was observed. Correspondingly, SEPs were abnormal for al1 four

lirnbs in patient 3. SEP abnomalities were more pronounced for the nght median and

posterior tibia1 nerves, in agreement with the severe structural abnormalities of the left

central region. Therefore, in these two, SEP abnormalities were only present if the

corresponding part of the post-central gyrus was found abnormal.

-

posteriorly located

N20iP20

- CSD:

-

ill-defïned N20P20

- A;a;CSD:

post. tib. -

1 N4W40

posteriorly located

post.tib. -

- CSD:

deflned N40/P40

- a;CSD: ill-

No such correlation was found for patient 2. SEPs were normal for all four limbs even

though a pensylvian polymicrogyria invaded the hand region on the right post-central

W S -

Modeied Generators of SEPs

Three-dipo le models explahed on average 92% of the variance in median nerve SEPs

while two-dipole models explained 93% of the variance in posterior tibia1 nerve SEPs.

Dipole pararneters obtained £kom patients were compared to the positions and orientations

obtained fiom control subjects. Positions and orientations which were more than 2 SD

away fiom the mean position and orientation of control subjects were considered

abnonnal.

Modeled Sozïrces us Location of Dysgen efic Lesions

A cornparison hetween the part of the post-central gyms invaded by the dysgenetic lesion

and abnormalities of the modeled sources was perfomed. In patient 1, al1 dipole

pararneters were within normal limits for the four limbs, in agreement with a lesion

located outside the post-central gynis. Figure 5 illustrates the modeled sources obtained

for this patient.

Fig 5. Modeled generators of SEPs in a patient with focal cortical dyspIasia located in the left superior parietal lobule. Dipole N20P20 models the generators of early cortical SEPs obtained following the stimulation of left and right median nerves. Dipole N40P40 models the generators of early cortical SEPs for the left and right posterior tibia1 nerves (axial slice on top image and sagittal slice on bottom image). After correction for eccentricity, al1 dipoles parameters were normal. In this patient, normal dipole parameters were observe in correlation with a dysgenetic Iesion Iocated outside the post-central gynis (black arrows).

Al1 sources identi5ed well the central sulcus (CS). On the sagittal plane illustrating

dipole N40/P40 for the nght postenor tibial nerve SEPS (nght lower corner), it can clearly

be seen that the modeled source is located anterior to the focal cortical dysplastic lesion

(black arrow). As was the case for control subjects, 2 sources (for the left ann and fcot)

occupied a very eccentric position outside the brain and radial projections of the dipoles

onto the surface of the brain were performed.

In patient 3, al1 sources presented abnomal parameters, in agreement with the bilateral

central lesions (figure 6).

Fig 6. Somatosensory representation of the 4 extremiies in a patient with bilateral assyrnetric dysgenetic lesions. The right central abnormal

pattern with no clear dysgenetic feature is associated with a very posterior source for the hand and a lateral source Dipole N50) for the legs. The posterior shifting of the Ieft hand is partly due to electrode misplacement. The severe dysgenetic lesion in the left central region explains the anterior displacement of somatosensory functions. The validity of the N40/P40 source obtained for the right leg is however questionable since only 67% of the variance was explained. Right leg

SEPs were raîher characterized by a low amplitude, possibly due to the proximity of the dysgenetic lesion-

Modeled sources presented a very diEerent pattern in the two hemispheres, which

correlated with the asymmetnc lesions (figure 6). In the right hemisphere, the tangentid

source modeling the N2OE20 became active around 20 ms and followed a normal

temporal evolution. However, dipole N20/P20 was located in the parietal lobe, 2.4 cm

fiom what appeared to be the central sulcus. The orientation of dipole N20/P20 formed a

79 O angle with the central sulcus, but it also formed a 78" angle with anterior wall of the

abnormal gyms on which it was located. The posterior localization of dipole N20/P20

following stimulation of the lei? median nerve is surprising, considering the parieto-

occipital abnormalities. Closer investigation showed that electrodes F8, T4, T6, F4 and

FC4 were systematically located posterior to their counterpart in the left hemisphere, at an

average distance of 1 1 mm. Since these electrode positions were registered one after the

other, it is most probable that the patient moved backwards with respect to the reference

Eame during registration of these electrodes and went back to his original position after.

Since electrode misplacernent was probably approximately 1 cm in the posterior direction,

this error does not completely explain why dipole N20/P20 was located a 24 mm posterior

to the central sulcus. Therefore, the posterior location of dipole N20/P20 for stimulation

of the left arm is thus equivocally abnormal.

For right posterior tibia1 nerve SEPs, dipole N40/P40 was located 2.4 cm deep to the

cortical surface and anterior to the central sulcus (more precise measurement could not be

made due to the depth of the dipole), but al1 other dipole parameters were nomal. The

second source, which modeled the N50 and P60 potentials in addition to dipole N40/P40,

was abnormal. In control subjects, dipole N50 usually occupies a central, parasagittal

location and, although the orientation is somewhat variable, it usually points anteriorly.

In patient 3, dipole N50 was of considerable importance in the 2-dipole mode1 as both

dipoles had fairly equal weight. Dipole N50 was located close to dipole N20/P20, on

what appeared to be the post-central gyms. Dipole N50 pointed laterally, upwards and

slightly posteriorly, and explained an isolated positivity maximally recorded at C4 around

49 ms post-stimulus. On a horizontal plane, the orientation of dipole N50 was almost

perpendicular to the orientation of the post-central gyrus.

For the lep henkphere, dipole N20P20 was located deep in the fiontal lobe and pointed

anteriorly and laterally to generate the large positivity maximally recorded at F3 and a

small negativity maxirnally recorded at CPz-CP2 at 22 ms. Dipole N20ff20, was located

slightly anterior to the dysgenetic lesion. For rÏght posterior tibial SEPs, 2-dipole rnodels

could explain only 67% of the variance in the data due to the srnail amplitude of the

potentials (generators model noise as well). This small field is not surprising considering

the large structural abnormality found in the left parasagittal central region. Although the

2-dipole model offers a poor explanation of the data due to the small signal-to noise ratio,

dipole N40P40 could be mapped onto MRI and was found anterior to the large

invagination, as was the case for dipole N20/P20 on the same side. These findings

suggest that somatosensory representation of the right hand and foot were displaced

anteriorly due to the central dysgenetic lesion.

In patient 2, abnormal dipole parameters were found for the left median nerve and nght

posterior tibia1 nerve SEPs. For lefl median nerve, dipole N20P20 was located 1 cm

posterior to the central sulcus, more than 2 SD fiom the average distance found in control

subjects. However, dipole N20/P20 was still located on the post-central gyms. For nght

posterior tibial nerve SEPs, dipole N40/P40 was located 1.6 cm parasagittally and 2.5 cm

deep to the cortical surface in the lefi hemisphere. Since all other dipole parameters were

normal and the latency, amplitude and CSD maps of SEPs presented no abnormalities, the

significance of these fuidings is difficult to assess. The apparent lack of correlation

between the presence of dysgenetic cortex in the hand region of the right post-central

g p s and normal SEPs could be explained by the location of the modeled source outside

the area of polymicrogyria visible on MRI (figure 3).

Fig. 7 Sensory representation of the left hand in a patient with right perisylvian poly- rnicrogyria. Normal SEPs were obtained following stimulation of the left rnedian nerve even though the dysgenetic lesion invaded the hand region on the nght post-central gyrus. The upper part of the lesion (white arrow) is located slightly under the modeied source. A possible explanation for the apparent discrepancy between structural and fûnctional data is that both the lesion and the modeled generator impinged on the hand region, but the generator was actually outside the lesion.

In this study, the location of modeled generators of SEPs was compared to the location of

the primary sornatosensory cortex, in the anterior wall of the post-central gyrus, to

evaluate how well source analysis maps somatosensory function in the normal human

brain. Prirnary somatosensory cortex was fomd by following published guidelines on how

to fïnd the central sulcus on CT and MR images (Kido et al 2980, Steinmetz et al 1990;

Naidich 199 1). Results indicated that, after correction for eccentricity, tangential sources

rnodeling the median nerve N20/P20 potentials could locate the centrai sulcus withul a

few millimeters. Dipole N40/P40 located the central sulcus with an average distance of

6.6 mm, but its location in the parasagittal region was more variable and tended to be

anterior to the central sulcus. Source analysis of rnedian and posterior tibia1 nerve can

thus i d e n t i ~ wih acceptable accuracy the primary somatosensory cortex on the anterior

part of the post-central gyms in neurologically normal subjects.

Other dipole parameters were analyzed to understand the charactenstics of the generators

of median and especially posterior tibia1 nerve SEPs. The acceptable varïability in dipole

parameters was documented in 10 neurologically normal individuals and clear patterns

ernerged. Dipole N20P20 pointed anteriorly and towards the midline, forming an average

angle of 75' (SD=lOO) with the post-central gyrus. Dipole N40P40 was somewhat more

variable in orientation, but it predominantly pointed towards the longitudinal fissure and

posteriorly with an orientation ranging fiom tangential to radial.

The study on control subjects also ailows the investigation of more theoretical questions.

The N20P20 and P22 potentials of median nerve SEPs were well modeled by 1

subcortical dipole and 2 cortical sources. The number of sources used to model recorded

wavefoms is always subject to discussion, but the 3-dipole model was supported by

previous research (Franssen et al. 1992; Buchner et al. 1994, 1995) and confirmed with

Principle Component Analysis in many cases. Three-dipole models usually consisted of

a tangentid+ radial parietal model and therefore approximated the tangential + somatosensory radial model proposed by AUison (Allison et al. 1980,1989). The radial

dipole, dipole P22, was usually located posterior to the post-central g p s in order to

model the transition between the two well defïned dipolar fields N20/P20 and N25/P25.

The poor fit obtained at electrodes FC3EC4 may suggest that the P22 potential was not

well modeled by dipole P22. However, a better definition of P22 based on a slightly

longer cortical interval was precluded by the be,-g of the N25P25 potentials.

Models were also biased towards one tangential rather than two radial dipoles to model

the N20/P20 because the high concentration of electrodes in the central region and the

lack of low electrodes favors the detection of tangential rather than radial dipoles.

Centrally located radial dipoles are best defined by low electrodes and electrodes at the

vertex. Tangentid dipoles are well defbed by a large array of electrodes on the top of the

head, which was the chosen montage in this study. Moreover, the beginning of cortical

intervals was defined by the apparition of the N20P20 potentials, excluding the isolated

fiontal positivity P20 seen in the previous time sample. The asynchronous apparition of

the N20 and P20 potentials speaks against a tangential dipolar model.

The experirnental paradigm used to model early cortical median nerve SEPs was similar

to protocols already pubrished (Buchner et al. 1995) wiîh the exception of the method used

to delineate the epoch containing the early cortical SEPs. However, the definition of the

epoch of interest is crucial since successive dipolar fields c m present slightly different

orientations md/or locations, which will result in interaction of the dipoles. For median

nerve SEPs, the orientation of instantaneous dipoles N20P20 and N25/P25 was

sometirnes separated by a large angle. A cortical interval which would include N25/P25

as well as N20P20 would contain 3 generators (with the radial dipole) very dose in space

and Likely to interact upon fitting the sources. Afthough delineation of intervals with

Global Field Power (GFP) is a more objective way of defining epochs of interest, GFP

cannot isolate desired generators when multiple sources are active very close in time.

Postenor tibial nerve SEPs can also be modeled by simple dipolar models, but 2 dipoles,

instead of 3, are generally sufficient to explain the data sets. Two-dipole models are

associated with a low residual variance in the cortical interval (6% on average).

Moreover, a horizontal or oblique dipole pointing towards the hemisphere ipsilateral to

stimulation makes physiological sense in view of the representation of the foot almg the

longitudinal fissure (Penfield and Boldrey 1937). The presence of a unique dipole to

model the N40/P40 potentials and additional dipole(s) at later latencies corroborates MEG

findings (Huttunen et al. 1987). It is difficult to evaluate the possibility of two radial

dipoles instead of a single tangntial dipole for the same reasons as was described for

median nerve SEPs. Even though variabiiity in the distributions of postenor tibial nerve

SEPs was reported in the literature (Emerson,1988), parameters describing dipole

N40P40 were quite stable across subjects.

The digitization of electrodes in MRI space provided a lïnk between functional data and

. the anatomical substrate and allowed direct comparisons. It has not been tested whether

digitization of electrodes greatly improves the localization of the generators inside the

spherical head model compared to the use of standard electrode coordinates, but the

representation of dipoles onto individual MEUS is surely superior to extrapolations nom

anatomical landmarks. Of course, registration of electrodes must be properly done as

mislocations will affect the position of the modeled sources.

Representation of dipoles in MRI space also allows for the cornparison of dipole

orientation with gyral patterns. In this study, cornparisons were made between the

orientation of dipole N20/P20 and the presumed generator of the N2Q/P20 potentials,

Brodmann's area 3b. The average angle of 75" (SD=lOO) separating dipole onentation

and the cIosest segment of the central sulcus is a good approximation of the 90" angle

expected on theoretical grounds. This idea could be explored M e r . Since orientation is

more accurately assessed than location in source analysis, onentation could contribute to

the mapping and solutions could be constrained to be normal to the cortical surface in the

context of realistic head models. A criterion of normality added to an automated search

for dipole location inside a realistically shaped head mode1 could significantly improve

source localization.

Dipole orientation c m also provide important information about the generators of SEPs.

Generators for posterior tibial nerves have not been well studied. The orientation of

dipole N40P40 is usually tangential, posterior, slightly upwards and approximately

parallel to the central sulcus, although radial dipoles are also observed. A radial dipole

could be due to a generator Iocated on the crest of paracentral lobule rather than in the

longitudinal fissure. The postenor and midline orientation of the tanpential dipole camot

be explained as easily. In hemispheres in which the central sulcus does not open up onto

the longitudinal fissure, it can be hypothesized that the geoerator is located at the most

medial end of the central sulcus. The fibers generating the SEPs would thus course dong

the central sulcus and generate the observed field. However, this small notch at the end of

the central sulcus is not present in cases in which the central sulcus opens up ont0 the

longitudinal fissure. This study did not compare the presence of the notch with the

orientation of dipole N40P40, but m e r investigation could address this issue.

The major pitfall of the study is the large eccenûicity of dipole N20@20, and occasionally

N40R40, in an important number of subjects. Although the same problem was observed

in patients and cornparisons were therefore possible, it is impossible to predict how the

error in the depth of the dipole affects the other spatial coordinates and the orientation of

the dipole. It was observed that a srnall energy criterion can sometirnes solve this

problem, although the mechanism for this effect is unclear. However, the use of variance,

energy or separation criteria represents considerable input fiom the investigator and c m

introduce biases in the solutions. In this study, a very systematic fitting strategy was used

to try to minimize biases. Variance, energy and separation were not used in this study and

fitting depended strictly on the reduction of residual variance. The problem of dipole

eccentricity will be M e r investigated to detennine i fa change in head mode1 will

reduce dipole eccentricity.

To conclude, the study on control subjects demonstrated that, when shortcomings such as

modeling errors in the depth of the dipoles are corrected for, source analysis of median

and posterior tibia1 nerve identifies, as expected, the primary somatosensory cortex on the

anterior part of the post-central gyrus. Since rnodelization of SEPs cm identiQ

somatosensory cortex in normal brains, it follows that it can also identiq somatosensory

function in abnomal brains, in which the post-central gyrus is sornetimes difficult to

identifjr (patient 3). The study on patients with dysgenetic lesions aimed at ascertaining

the representation of the normal sornatosensory functions observed in these patients. It is

of interest to determine whether normal sensory funciion is sustained by the dysgenetic

cortex, by surrounding nomal cortex of by subcortical systems.

Models obtained in the patients group were associated with a residual variance

comparable to that found in control subjects. Models made physiological sense as, in al1

cases, dipole N20/P20 showed an orientation approximately nomal to the post-central

gyms or the gyms on which dipole N20P20 was located. The oniy exception was the

generator of nght hand SEPs in one subject which was displaced to the centnun ovale of

the fiontal lobe by a large left centrd alolding and was thus f a fkom gyral/suIcaI

patterns.

Models were mapped in MRI space and compared to the Location of the lesion and the

central sulcus, when such an anatomical Iandrnark was identifiable. Ln patients 1 and 2,

modeled generators represented al1 somatosensory fünctions in close proximity to the

central sulcus and outside dysgenetic lesions. Ln patient 3, the postenor displacement of

the left hand representation was probably partly attributable to electrode misplacement,

and partly due to an atypical mapping of function. The left leg was represented in the

same area but slightly anterior to the left arm representation (on the post-central gyrus), in

addition to the usual right central, parasagittal position. The reason for these shifts in

position is not clear, as the whole region presents ,VI abnormalities, without clear

dysgenetic features. The somatosensory representation of the right hand was displaced to

the non-lesionai fiontal lobe, due to the presence of a severe left central dysgenetic lesion.

In sumrnary, al1 SEP generators for which adequate models could be obtained were

located outside dysgenetic cortex, with the possible exception of the representation of the

left hand of patient 3 in parieto-occipital polymicrogyria.

Abnormal potentials recorded outside dysgenetic cortex can be attributed to more

widespread dysgenesis invisible on MRI. Rowever, the distribution of SEPs is also

f ict ion of _=al anatomy and it is quite possible that the generators of displaced

potentials present a different anatomical configuration, resulting in different fields.

In patient 3, the amplitude of SEPs obtained following stimulation of the right leg was so

small that modeled generators could only account for 67% of the variance in the data,

because the mode1 was explaining noise. An absence of potentials or srna11 amplitude

potentials were aIso observed in previous studies (de Rijk-van Andel et al. 1992; Di Capua

et al. 1993; Vossler et al. 1992; Raymond et al. 1997). It is possible that the lack of early

cortical potentials observed for right leg SEPs in patient 3 resulted from the activity of

poorly aligned pyramidal cells located in the dysgenetic lesion. Pyramidal cells are the

presumed generators of evoked potentials and the electrical field produced by these cells

is a fùnction of their alignment and the synchronicity of the signal. It is quite unliicely that

a 32 electrode coverage of the head would have "missed" a displaced potential of normal

amplitude. The modeled generator which explained oniy 67% of the data was located

anterior to the central lesion, as was the case for dipole N20P20 on the sarne side. One

c m thus infer an anterior displacement of the sensory region, but the representation of the

leg may border on the dysgenetic lesion, resulting in smaller potentials.

To conclude, it is quite feasible to apply the experimental paradigm designed for control

subjects to the study of patients wiîh dysgenetic lesions. As was seen, the uiterpretation

of SEPs or modeled generators is often hùidered by the lack of fundamental research in

this area. The fact that only the very early potentials were mapped also Limits the

interpretation, as later potentials (eg. in the 30-45 ms range for median nerve SEPs) could

well be representative of the localization of somatosensory functions. In addition, one c m

foresee problems with electrode digitkation in the context of non-cooperative patients.

Digitization of electrodes could be improved by comparing electrode positions to a

chosen reference point on the head, so that if the patient moved, the reference frame

would consequently rnove. Finally, it is quite possible that fields observed in cases of

severe dysgenetic abnomalities will not adopt a dipolar configuration, rendering dipole

source analysis an inadequate modeling tool.

Further investigation with a Iarger number of subjects should identi* how different types

of cortical dysgenesis affect SEPs. The effect of different head models and of different

modeling procedures on the eccentricity of the source should be carefiilly investigated.

An automated search for sources with an orientation normal to the cortex in realistically

shaped head mode1 could result in significantly improved Iocalization of function.

Abbnizzesse, G., Dall'Agata, D., Morena, M., Reni, L., Favale, E. Abnomalities of

Parietal and PreroIandic Somatosensory Evoked Potentials in Huntington's Disease.

Electroenceph cIin Neurophysiol, 1990, 77: 340-346.

Allison, T., Goff, W.R., Williamson, P.D-, Van Gilder, J-C. On the Neural Origin of Early

Components of the Huma. Somatosensory Evoked Potential. In: Chical Uses of

Cerebral, Brainstem and Spinal Samatosensur y Evoked Potentials. Progress in Clinical

NeuuophysioZo~y. Edited by JE Desmedt. Basel, Karger, 1980: 5 1-68.

Allison, T., Wood, C.C., Goff, W.R. Brain Stem Auditory, Pattern-Reversai Visual, and

Short-Latency Somatosensory Evoked Potentials: Latencies in Relation to Age, Sex and

Brain and Body Size. Electroenceph ciin Neurophysiol, l983,55: 619-636.

Allison, T., McCarthy, G., Wood, CC., Darcey, T.M., Spencer, D.D., Williamson, PD.

Human Cortical Potentials Evoked by Stimulation of the Median Nerve. 1.

Cytoarchitectonic Areas Generating S hort-Latency Activity . J Neurop hysiol, 2 989,62:

694-710.

Aminoff, ML, Eisen, A., Somatosensory Evoked Potentials. In EZectrodiagnosis in

Chical Neurology. Edited by M.J. Amino& New York: Churchill Livingstone, 1996:

571-603.

Barkovich, A.J., Guerrini, R., Battaglia, G., Kalifa, G., NGuyen, T., Panneggiani, A.,

Santucci, M., Giovanardi-Rossi, P., Granata, T., D'Incerti, L. Band Heterotopia:

Correlation of Outcome with Magnetic Resonance Irnaging Parameters. Annals of

Neurology, 1994,36: 609-617.

Baumgartner, C., DoppeIbauer, A., Sutherling, W.W., Lindinger, G., Levesque, M.F.,

AuU, S., Zeitlhofer, J., Deecke, L. Somatotopy of Human Hand Somatosensory Cortex as

Studied in Scalp EEG. Electroenceph clin Neurophysiol, 1993, 88: 271-279.

Brazier, M.A.B. A study of the electrical field at the surface of the head. Electroenceph

clin Neurophysiol, 1949,2(Suppf): 3 8-52.

Brenner, D., Lipton, J., Kauhan, L., Williamson, S.J. Somatically evoked magnetic

fields of the human brain. Science, 1978, 199: 81-83.

Broughton, R., Discussion. In: Average Evoked Potentials: Methods. Results, and

Evalirations. Edited by E Donchin, DB Lindsley. Washington: National Aeronautics and

Space Administration, 1969: 79-84.

Broughton, R.J. Somatosensory Evoked Potentials in Man: Cortical and Scalp

Recordings. McGill University, Montreal, Quebec, 1967.

Buchner, H., Adams, L., Knepper, A., Ruger, R., Laborde, G., Gilsbach, J.M., Ludwig, I.,

Reul, J., Scherg, M. Preoperative localization of the central sulcus by dipole source

analysis of early somatosensory evoked potentids and three-dimensional magnetic

resonance irnaging. J Neuroswg, 1994,80: 849-856.

Buchner, H., Adams, L., Muller, A., Ludwig, L, Knepper, A., Thron, A., Niemann, K.,

Scherg, M. Somatotopy of Human Hand Somatosensory Cortex Revealed by Dipole

Source Analysis of Early Somatosensory Evoked Potentials and 3D-NMR Tomography.

Electroenceph clin Neurophysiol, 1995,96: 12 1-1 34.

Carpenter, M.B. Core Text of Neuroanatomy. Baltimore: Williams and Wilkins, 199 1 :

268.

Chiappa, K.H., S hort-Latency Somatosensory Evoked Po tentials : Methodology. In:

Evoked Potentiak in Chical Medicine. Edited by KH Chiappa. New York: Raven Press

Ltd, 1989: 307-371.

Chu, N. Somatosensory Evoked Potentials: Correlations with Height. Electroenceph clin

Neurophysiol, 1986,65: 276.

Cohen, L.G., Starr, A. Locaiization, Timing and Specificity of Gating of Somatosensory

Evoked Potentials during Active Movement in Man. Brain, l987,llO: 45 1-467.

Cohen, M.S., Bookheimer, S.Y. Localization of Brain Function Using Magnetic

Resonance Imaging. Trends in Neurosciences, 2994, 2 7: 268-277.

Cuffin, B.N., Cohen, D., Yunokuchi, K.e. Tests of EEG Localization Accuracy Using

Implanted Sources in the Human Brain. A m Neurol, 199 1,29: 132-1 38.

Dawson, G.D. Cerebral Responses to Nerve Stimulation in Man. Brain Medical Bulletin,

1950,6: 326.

de Rijk-van Andel, J.F., Arts, W.F.M., de Weerd, A.W. EEG and Evoked Potentials in a

Series of 2 1 Patients with Lissencephaly Type 1. Neuropediatrics, 1992,23: 4-9.

Deiber, M.P., Giard, M.H., Mauguiere, F. Separate Generators with Distinct Orientations

for N20 and P22 Somatosensory Evoked Potentials to Finger Stimulation? Electroenceph

clin Neurophysiol, 1986,65: 321-334.

Delestre, F., Longchampt, P., Dubas, F. Neural Generator of Pl4 Far-Field

Somatosensory Evoked Potentialstudied in a Patient with Pontine Lesion. Electroenceph

clin Neurophysiol, l986,65: 227-230.

Desmedt, LE., Nguyen, T.H., Bourguet, M. Bit-Mapped Color Imaging of Human Evoked

Potentials with Reference to the N20, P22, P27 and N30 Somatosensory Responses.

Electroenceph clin Neurophysiol, l987,68: 1-1 9.

Desmedt, J.E., Cheron, G. Central Somatosensory Conduction in Man: Neural

Generators and hterpeak Latencies of the Far-Field Components Recorded fiom Neck

and Right or Left Scaip and Earlobes. Electroenceph clin Neurophysiol, 1980% 50: 382-

403.

Desmedt, J.E., Cheron, G. Somatosensory Evoked Potentids to Finger Stimulation in

Healthy Octogenarians and in Young Adults: Wave Forms, Scalp Topography and

Transit Times of Parietal and Frontal Components. Electroenceph clin Neurophysiol,

1980b, 50: 404-425.

Desmedt, J.E., Cheron, G. Non-Cephalic Reference Recording of Early Somatosensory

Potentials to Finger Stimulation in Adult or Aging Normal Man. Differentiation of

Widespread N18 and Contralateral N20 f?om the Prerolandic P22 and N30 Components.

EIectroenceph clin Neurophysiol, 198 1,52: 553-570.

Di Capua, M., Vigevano, F., Wisniewski, K. Somatosensory Evoked Potentials in

Hemimegalencephaly and Lissencephaly: Anatomo-Functional Correlations. Brain and

Development, 1993, 15: 253-257.

Emerson, R.G. Anatomic and Physiologie Bases of Posterior Tibia1 Nerve Somatosensory

Evoked Potentials. Neurologie Clinics, l988,6: 735-749.

Fox, P.T., Burton, H., Raichle, M.E. Mapping Human Somatosensory Cortex with

Positron Emission Tomography. Journal of Neurosurgery, 1987,67: 34-43.

Franssen, H., Stegeman, D.F., Moleman, J., Schoobaar, R.P. Dipole Modeling of Median

Nerve SEPs in Normal Subjects and Patients with Srnall Subcortical Marcts.

Electroenceph clin Neurophysiol, 1992, 84: 40 1-41 7.

Freeman, W.J. Mass Action of the Newous System. New York: Academic Press, 1975.

Green, J.B., Walcoff, M., Lucke, J.F. Phenytoin Prolongs Far-Field Somatosensory and

Auditory Evoked Potential Interpeak Latencies. Neurology, 1982a: 85-88.

Green, J.B., Walcoff, M., Lucke, J.F. Cornparison of Phenytoin and Phenobarbital Effects

on Far-Field Auditory and Somatosensory Evoked Potential Interpeak Latencies.

Epilepsia, l982b: 41 7-42 1.

Halonen, J.P., Jones, S., Shawkat, F. Contribution of Cutaneous and Muscle Afferent

Fibres to Cortical SEPs Following Median and Radial Nerve Stimulation in Man.

Electroenceph clin Neurophysiol, 2988,71: 33 1.

Hammeke, T.A., Yetkin, F.Z., Wade, M.M., Moms, G.L., Haughton, V.M., Rao, SM.,

Binder, J.R. Functional Magnetic Resonance Imaa$ng of Somatosensory Stimulation.

Neurosurgery, 2 994,3 5 : 677-68 1.

He, B ., Musha, T. Effects of Cavities on EEG Dipole Localization and their Relations

with Surface Electrode Positions. Int J Biomed Comput, 1989,24: 269-282.

Helmholtz, G. über einige Gesetze der Verteilung elektnscher Strome in korperlichen

Leitem, mit Anwendung auf die thierischelektrische Versuche. Ann Phys Chem, 1853 :

Hume, A.L., Cant, B.R., Shaw, N.A. Central Somatosensory CT in Cornatose Patients.

Annals of Neurology, 1979,5: 379-84.

Hume, A.L., Cant, B.R., Shaw, N.A., Cowan, J-C. Central Somatosensory Conduction

Time fiom 10 to 79 Years. Electroenceph clin Neurophysiol, 1982,54: 49-54.

Hume, A.L., Cant, B.R. Conduction Time in Central Somatosensory Pathways in Man.

Electroenceph clin Neurophysiol, 1978,45: 361-375.

Huttunen, J., Kaukoranta, E., Hari, R. Cerebral magnetic responses to stimulation of tibial

and sural nerves. J Neurol Sci, 1987,79: 43-54.

Kakigi, R. The Effect of Aging on Somatosensory Evoked Potentials Following

Stimulation of the Posterior Tibial Nerve in Man. Electroenceph clin Neurophysiol, 1987,

15: 277-286.

Kakigi, R., Shibasaki, H. Scalp topography of the short latency somatosensory evoked

potentials following posterior tibial nerve stimulation in man. Electroenceph clin

Neurophysiol, 1983, 56: 430-437.

Kakigi, R., Shibasaki, H. Effects of Age, Gender and Stimulus Side on Scalp Topography

of Somatosensory Evoked Potentials Following Median Nerve Stimulation. Journal of

Clinical Neurophysiology, 199 1, 8: 320-330.

Kakigi, R., Shibasaki, H. Effects of Age, Gender, and Stimulus Side on the Scalp

Topography of Somatosensory Evoked Potentials Following Postenor Tibial Nerve

Stimulation. Journal of Clinical Neurophysiology, l992,g: 43 1-440.

Kawamura, T., Nakasato, N., Seki K., Shimizu, H., Fujiwara, S., Yoshimoto, T. Locating

the Central Sulcus using Three Dimensional Image Fusion S ystem of Helmet Shaped

Whole Head MEG and MRI (Japanese). No Shinkei Geka - Neurological Surgery, 1994,

Kawamura, T., Nakasato, N., Seki K., Kanno, A., Fujita, S., Fujiwara, S., Yoshimoto, T.

1996, Neuromagnetic evidence of pre-and post-central cortical sources of somatosensoq

evoked responses. Electroenceph clin Neurophysiol, 100: 44-50.

Kido, D.K., Le May, M., Levinson, A. W., B enson, W.E. Computed tomographie

localizaîion of the precentral gyrus. Radiology, 1980, 135: 373-377.

Kimura, J., Mitsudorne, A., Beck, D.O., Yamada, T., Dickins, Q.S. Field Distribution of

Antidrornically Activated Digital Nerve Potentials: Mode1 for Far-Field Recording.

Neurol0~7, t 983, 33: 1 164-1 169.

Kimura., J., Mitsudome, A., Yamada, T., Dickins, Q.S. Stationary Peaks fkom a Moving

Source in Far-Field Recording. Electroenceph clin Neurophysiol, 1 984,S8 : 3 5 1 -3 6 1.

Kitamura, K., Kamio, Y., Futamata, H., Hashimoto, T. Tibial Nerve Somatosensory

Evoked Potentials: Studies on Latency Variability as a Function of Subject Height and

Age. Rïnsho Byori - Japanese Journal of Clinical Pathology, 1995,43: 81-86.

Kuzniecky, R., Cascino, G.D., Palmini, A. Structural irnaging. In: Surgical treatment of

the epilepsies. Edited by JJ Engel. New York: Raven Press, 1993: 197-100.

Landau, W.M., Evoked Potentials- In: The Neurosciences - A Study Program. Edited b y

GC Quarton, T Melnechuk, FO Schmitt. New York: Rockefeller University Press, 1 967.

Lastimosa, A.C.B., Bass, N.H., Stanback, K., Norvell, E.E. Lumbar Spinal Cord and

Early Cortical Evoked Potentials After Tibial Nerve Stimulation: Effects of Stature on

Normative Data. Electroenceph clin Neurophysiol, 1982, 54: 499-507.

Mauguiere, F., Desmedt, J.E., Coujon, J. Astereognosis and Dissociated Loss of Frontal

or Parietal Components of Somatosensory Evoked PotentÏaIs in Hemisphenc Lesions.

Brain, 1983, 106: 271-3 1 1.

Mauguiere, F., Ibanez, V. Loss of Parietal and Frontal Somatosensory Evoked Potentials

in Hemispheric Deaftèrentation. In: Nao Trends and Advanced Techniques in Chical

NettrophysioI0g-y. Edited by PM Rossini, F Mauguiere. Amsterdam: Elsevier, 1990: 274-

285.

Mervaala, E., Paakkonen, A., Partanen, J.V. The Wuence of Height, Age and Gender on

the Iriterpretation of Median Nerve SEPS. Electroenceph clin Neurophysiol, 1988, 71:

109-1 13.

Naidich, T.P. MR imaging of brain surface anatomy. Neuroradiology, 199 1,33(Suppl):

S95-S99.

Nogueira, M C , Bninko, E., Vandesteen, A., De Rood, M., Zegers de Beyl, D.

Differential Effects of Isoflurane on SEP recorded over Parietal and Frontal Scalp.

Neurology, 1989,39: 1210-2215.

Papakostopoulos, D., Crow, H.J. Direct Recording of the Somatosensory Evoked

Potentials fiom the CerebraI Cortex of Man and the Difference between Precentral and

Postcentral Potentials. In: Brainstem and Spinal Somatosensory Evoked Potentials.

Progress in Chical Neurophysio2og-y. Edited by JE Desmedt. Basel: Karger, 1980: 15-

26.

Penfield, W., Boldrey, E. Somatic Motor and Sensory Representation in the Cerebral

Cortex of Man as Studied by Electrical Stimulation. Br&, 1937, 60: 389-443.

Raymond, A.A., Jones, S.J., Fish, D.R., Stewart, J., Stevens, J.M. Somatosensory evoked

potentials in adults with cortical dysgenesis and epilepsy. Electroenceph clin

Neurophysiol, 1997, 104: 132-142.

Romani, A., Bergarnaschi, R., Versino, M., Delnevo, L., Callieco, R,, Cosi, V. Spinal and

Cortical Potentials Evoked by Tibia1 Nerve Stimulation in Humans: Effects of Sex, Age

and Height. Bolletîino - Societa Italiana Biologia Sperimentale, l992,68: 69 1-698.

Scherg, M. Fundamentals of Dipole Source Potential Analysis. In: Auditov Evoked

Magnetic Fields and Elech-ic Poten tials. Edited by F Grandori, M Hoke, GL Romani.

Basel, Karger, 1990: 40-69.

Scherg, M., Ebersole, J.S. Models of Brain Sources. Brain Topogr, 1993, 5 : 41 9-423.

Scherg, M., von Cramon, D. Two Bilateral Sources of the Late AEP as Identified by a

Spatio-Temporal Dipole Model. Electroenceph clin Neurophysiol, 1985,62: 32-44.

Scherg, M., von Cramon, D. Evoked Dipole Source Potentials of the Human Auditory

Cortex. Electroenceph clin Neurophysiol, 1986, 65: 344-360.

Seyal, M., Emerson, R.G., Pedley, T.A. Spinal and early scalp-recorded components of

the somatosensory evoked potential following stimulation of the postenor tibia1 nerve.

Electroenceph clin Neurophysiol, l983,55: 320-330.

Smith, D.B., Sidman, RB., Henke, J.S., Flanigin, H., Labiner, D., Evans, C.N. Scalp and

Depth Recordings of Induced Deep Cerebral Potentials. Electroenceph clin Neurophysiol,

1983,5: 145-150.

Snyder, A.Z. Dipole Source Localization in the Study of EP Generators: a Critique.

Electroenceph clin Neurophysiol, 199 2,80: 32 1-325.

S teinmetz, H., Furst, G., Freund, H. J. Variation of pensylvian and calcarine anatomic

landmarks within stereotaxic proportional coordinates. AJNR: Am J Neuroradio, 1990,

11: 1123-1 130

Vossler, D.G., W i b s , R.J., Pilcher, WH., Farwell, J.R. Epilepsy in Schizencephaly:

Abnormal Cortical Organization Studied by Somatosensory Evoked Potenhals. Epilepsia,

1992, 33 : 487-494.

Woolsey, C.N. Organization of Somatic Sensory and Motor Areas of the Cerebral Cortex.

In: Biological and Biochernical Bases ofBehavior. Edited by HF Harlow, CN Woolsey.

Wisconsin: University of Wilconsin Press, 19%: 63-8 1.

Yamada, T., Matsubara, M., Shiraishi, G., Yeh, M., Kawasaki, M. Topographie analyses

of somatosensory evoked potentials following stimulation of tibial, sural and lateral

femoral cutaneous nerves. Electroenceph clin Neurophysiol, 1996, 100: 33-43.

Yang, T.T., Gallen, C C , Schwartz, B.J., Bloom, F.E. Noninvasive Somatosensory

Homonculus Mapping in Humans by Using a Large-Array Biornagnetometer. National

Academy of Sciences of the United States of Pmerica, 1993, 90: 3098-3 102.

Zegeis de Beyl, D., Delberghe, X., Herbaut, A-G., Bruriko, E. The Sornatosensory Central

Conduction Tirne: Physiological Considerations and Normative Data. Electroenceph clin

Neurophysiol, 1988, 171: 17-26-

IMAGE EVALUATION TEST TARGET (QA-3)

APPLIED IMAGE, lnc 1653 East Main Street - -. - - Rochester. NY 14609 USA -- --= Phone: 71 6/482-0300 -- -- - - FX 71 W88-5989

O 1993. Applied Image. Inc., Al1 Rights Reserved