Journal of Neuroscience Methods 93 (1999) 163–168

Two-dimensional photon counting imaging and spatiotemporalcharacterization of ultraweak photon emission from a rat’s brain

in vivo

Masaki Kobayashi a,*, Motohiro Takeda a, Ken-Ichi Ito b, Hiroshi Kato b,Humio Inaba a,c

a Biophotonics Information Laboratories, Yamagata Ad6anced Technology Research and De6elopment Center, 2-2-1 Matsuei,Yamagata 990-2473, Japan

b Department of Physiology, Yamagata Uni6ersity School of Medicine, 2-2-2 Iida-nishi, Yamagata 990-9585, Japanc Tohoku Institute of Technology, 35-1 Yagiyama-kasumi-cho, Taihaku-ku, Sendai 982-8577, Japan

Received 19 January 1999; received in revised form 27 August 1999; accepted 2 September 1999

Abstract

The process of metabolic reactions within living cells leads to spontaneous ultraweak light emission. The development of asystem for highly sensitive imaging and spatiotemporal analysis of ultraweak photon emission from a rat’s brain is reported in thispaper. The equipment used in this experiment consists of a two-dimensional photon-counting tube with a photocathode measuring40 mm in diameter, a highly efficient lens system, and an electronic device to record time series of a photoelectron train withspatial information. The sensitivity and ability to extract spatiotemporal information from sequential data of a single photoelec-tron train were examined. The minimum detectable radiant flux density of the system was experimentally estimated to be9.9×10−17 W/cm2 with a 1-s observation time. Spontaneous photon emission was demonstrated from an exposed rat’s cortex invivo without adding any chemical agent or employing external excitation. An image of ultraweak photon emission was comparedwith one obtained after cardiac arrest. The intensity after cardiac arrest was depressed to approximately 60% of before that. Theregional properties of time courses of emission intensity were also demonstrated, indicating the potential usefulness forspatiotemporal characterization of photon emission with mapping of physiological information such as oxidative stress. Thistechnology constitutes a novel method, with the potential to extract pathophysiological information from the central nervoussystem. © 1999 Elsevier Science B.V. All rights reserved.

Keywords: Biophoton; Imaging; Photon counting; Reactive oxygen; Oxidative stress; Chemiluminescence

www.elsevier.com/locate/jneumeth

1. Introduction

During metabolic reaction processes, living organ-isms spontaneously emit ultraweak light without anyexternal excitation or administration of chemilumines-cent agents. This is referred to as ‘biophoton emission’(Cilent, 1988; Inaba, 1988; Popp, 1988; Popp et al.,1988; Slawinski, 1988; Van Wijk and Schamhart, 1988;Usa et al., 1994; Amano et al., 1995; Devaraj et al.,1997).

The intensity of emission is estimated to be less than10−16 W/cm2 on the surface. The biochemical mecha-nism by which biophotons are emitted is not dependenton specific enzymes or proteins for luminescence; so itcan be differentiated from bioluminescence phenomenaobserved, for example, in fireflies or luminescent bacte-ria, which have distinctive mechanisms for providingluminescence with a high quantum yield that can berecognized by the human eye.

Electronically excited states of constituents of livingcells are generally derived from oxidative metabolismthat accompanies the production of reactive oxygenspecies (ROS; O−

2� , H2O2, OH�, 1O2). In normal energymetabolism, cellular respiration, a reaction in the elec-tron transfer chain of inner mitochondrial membranes,

* Corresponding author. Present address: Yamagata TechnopolisFoundation, Yamagata Advanced Technology Research and Devel-opment Center, 2-2-1 Matsuei, Yamagata 990-2473, Japan. Tel.:+81-23-647-3130; fax: +81-23-647-3109.

E-mail address: kob@ckk.ymgt-techno.or.jp (M. Kobayashi)

0165-0270/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 5 -0270 (99 )00140 -5

M. Kobayashi et al. / Journal of Neuroscience Methods 93 (1999) 163–168164

participates in ROS production, which is especiallyfacilitated under the highly reduced state of an electrontransfer chain. Biophoton emission reflects the patho-physiological state with respect to energy (ATP) pro-duction by inner mitochondria and susceptibility tooxidative stress, which is caused by excessive produc-tion of ROS or a lack of activity for antioxidantprotection (Boveris et al., 1980; Cadenas et al., 1984).Adamo et al. (1989) showed that triiodothyronine-in-duced hypermetabolism in an in vivo rat’s brain leadsto increased ultraweak photon emission as a reflectionof oxidative stress and suggested that ROS contributedto premature aging in these animals.

We have described a highly sensitive imaging tech-nique that is based on a two-dimensional photon detec-tion and spatiotemporal characterization of ultraweaklight (Kobayashi et al., 1996, 1997a). To accomplishthis, we made the system based on a technique asconsecutive measurements of spatial and temporal in-formation of individual photoelectrons with adequateresolution of time. It also provides a photon correlationfunction in two-dimensional field with variety of timeresolution, and could contribute to the extraction ofnon-random signals connected with physiological infor-mation. This method has an advantage of capability onanalysis with varying time resolution and spatial resolu-tion. However, it is not available in commercial equip-ment of two-dimensional photon counting systems,because of the restricted time resolution under thetime-lapse mode of imaging, and these systems alsohave a disadvantage with a small area of photocathodefor macroscopic imaging such as our application.

The aim was to establish a technique whereby patho-physiological information can be visualized in vivo(Kobayashi et al., 1997b). In the current paper, thissystem, by which biophoton images of a rat’s braincould be characterized, was evaluated.

2. Materials and methods

2.1. Two-dimensional photon counting apparatus andanalytical procedures

The imaging system consisted of a two-dimensionalphoton counting tube with a large active area, a highlyefficient lens system installed in a sample chamber, andelectronic equipment for identifying two-dimensionalspatial and temporal photoelectron data (Kobayashi etal., 1996). A block diagram is shown in Fig. 1. Thephoton counting tube (Model IPD 440, Photek, UK)has a photocathode measuring 40 mm in diameter, witha spectral sensitivity (S-20). It operates at a wavelengththat ranges from 350 to 900 nm and with a quantumefficiency of 9% at 500 nm, 5.5% at 600 nm, and 1.3%at 800 nm. The active area of the photocathode forimaging is a 25×25 mm square.

The tube dark count was 76/s over the whole effec-tive area, with a thermoelectronic device for cooling at−35°C. Spatial resolution of the tube, which was deter-mined by the readout precision of the resistive anodeincorporated into the photon counting tube, was ap-proximately 200 mm.

Fig. 1. A block diagram of a highly sensitive imaging and spatiotemporal analysis system for ultraweak photon emission. Timing pulses inphotoelectron event from a position computer operate a four-cascaded counter to measure pulse-to-pulse interval based on an 80-MHz clock.Time interval data and position data are transferred to a buffer memory, which is operated alternately for data storage and transfer, via FIFO(fast-in-fast-out) memories, and finally are stored on a hard disk. IF represents an interface of each block of the circuit.

M. Kobayashi et al. / Journal of Neuroscience Methods 93 (1999) 163–168 165

Fig. 2. Operational timing chart of four cascaded counters in a time interval counter.

A specially designed lens system (Fuji Optical Co.,Tokyo, Japan) has a 0.45 NA (numerical aperture) andis composed of six lenses and a built-in shutter mecha-nism. Magnification of the lens system is 1.0, whichcorresponds to an image size of 25×25 mm. Relativeillumination of the lens system at an image height of12.5 mm from the light axis is 88% on the axis. Spatialresolution of the lens system was designed for approxi-mately 5 line-pairs/mm to provide a superior NA.

Output pulses from the resistive anode are fed to aposition computer (IPD controller, Photek, UK) todetermine the X–Y position of each photoelectronevent. These data (nine bits, two channels) are consecu-tively transferred to a specially designed pulse intervalcounter. Simultaneously, it stores 27 bits of timingdata, represented as the time interval between twosuccessive photoelectron pulses, with two-dimensionalposition data of each photoelectron event. The acquisi-tion circuit for timing data is composed of a four-cas-caded counter. Each stage of the counter is operated byphotoelectron pulses that serve as start and stop trig-gers in turn, as illustrated on the timing chart of Fig. 2.The cascaded structure of the counter potentially ac-cepts successive pulses with a period of 37.5 ns withouta miss-count. However, the time resolution of dataacquisition is restricted by the dead time of the positioncomputer to a pulse-pair resolution of 10 ms. The timeresolution of the system can be improved if the calcula-tion rate for each photoelectron on the position com-puter can be increased. The maximum time interval forphotoelectrons, which is determined by the clock fre-quency and data length of the counter (27 bits), is 1.7 swith an 80-MHz clock. The miss-count probabilitycaused by a counter overflow is theoretically negligible,even in the case of dark counts assuming a randomtime series. The total data record length depends on the

hard disk capacity of the computer. Here, using a 1Gbyte hard disk, it is 1.25×108 events.

After data acquisition, the data are transferred to aworkstation (Model SUN 3, Sun Microsystems, CA) toreconstruct the photon counting images and analyzespatiotemporal properties. This process demonstratesthe intensity kinetics in the regions of interest or thetime–space correlation of the photoelectrons.

2.2. The process of data collection for spatiotemporalcharacterization and photon correlation analysis

Raw data accumulated in the pulse interval counterare expressed by a set of sequences {(x1, j1), (x2, j2),…,(xN, jN)} where ji is the arrival time interval betweenthe (i−1)th and the ith photoelectron and xi is thetwo-dimensional location of the ith photoelectron; andN is the total number of photoelectron pulses. Initially,a photon counting image, expressed as n(x, y), which isthe number of photoelectron pulses at position (x, y)during a total measurement period, is constructed. Af-ter setting regions of interest (rj={[xj1, xj2], [yj1, yj2]})on the image plane, the number of photoelectron pulsesper unit time T (observation time) in each region arecalculated as n(rj, t, T) to extract the intensity timecourse. The spatial and temporal correlations of emis-sion intensity �I(r1, t) I(r2, t+t)� are calculated fromthe photoelectron correlation that is derived from thephotoelectron detection probability at a space-timepoint (r1, t) and detection probability after time t at r2,represented by the conditional probability Pc(r2, t) oftwo successive photoelectron pulses. The relationship isexpressed as

�I(r1, t) I(r2, t+t)�=a�n(r1, t, T)�Pc(r2, t)

M. Kobayashi et al. / Journal of Neuroscience Methods 93 (1999) 163–168166

where the brackets � � denote ensemble averaging anda, a constant.

The advantage of this system is the arbitrary selectionof spatial and temporal dimensions (rj and T, t), whichcan be achieved from a single measurement data pointinstead of making a fresh measurement for each changein the spatial or temporal dimensions (Kobayashi et al.,1996).

2.3. General preparation of animals and measurementprocedure for spontaneous photon emissions from theirbrains

Male Wistar rats (weighing 220–300 g) were anes-thetized in accordance with the NIH Guide for the Careand Use of Laboratory Animals. The animals weretracheotomized under anesthesia with pentobarbital (40mg/kg body weight, i.p.) and artificially ventilated aftercannulation of the external jugular vein for infusion(rodent respirator Model 683, Harvard, MA) with a gasmixture of 1% isoflurane in 45% oxygen and 55%nitrogen. The rats were immobilized by pancuroniumbromide (1 mg/kg, i.v.) and an additional dose (1 mg/kgper h) was subsequently infused with 1 ml/h of saline,using a syringe pump (Model TOP-5200, TOP Co.,Tokyo, Japan).

The rats were surgically prepared with an incision ofthe skin to expose the skull. The parietal bone wasremoved bilaterally while preserving the dura mater.Rectal temperature was monitored throughout the ex-periment and maintained at 38.290.5°C by using a hotwater blanket. Each rat was mounted on a stereotaxicframe and placed under the lens system in a completelylight-tight chamber. After focusing with weak lightillumination and dark adaptation to reduce phosphores-cence (1 h), biophoton measurement was started.

3. System performance

3.1. Minimum detectable radiant flux density

Minimum detectable optical power is determined byshot noise of the detector dark count. When minimumdetectable optical power is defined as unity of thesignal-to-noise ratio, the minimum detectable radiantflux density on the sample surface (Pmin; W/cm2) isexpressed as

Pmin=hcl

sd

h(V/2p)TA

with consideration of solid angle (V) for the lightcollection efficiency of the lens system. Here h is thequantum efficiency of the photocathode at wavelength l

and sd the standard deviation of the dark count withinthe 1-s observation time; T is the observation time for

integration and A is the active area of the detector; c andh represent the velocity of the light and Plank’s constant,respectively.

sd was obtained experimentally to be 8.9 counts/s. Pmin

at an observation time T=1 s, was estimated to be8.0×10−17 (W/cm2) (=2.5×102 photons/s per cm2 at630 nm) using the parameters h=5.5% and V/2p=10.2%,both of which were estimated from the numericalaperture of the lens system. This means that a singlephotoelectron corresponds to 1.8×102 photons on thesample surface.

The above was evaluated experimentally by using auniform light source, incorporated with an electrolu-minescent sheet (30×30 mm) and neutral density filters.The peak wavelength of the spectrum was 630 nm witha 120 nm width. Intensity was calibrated by anotherphoton counting system (Kobayashi et al., 1998) to be1.4×105 photons/s per cm2. When the light source wasinstalled in the focal plane, the averaged total photo-count was 4.1×103 counts/s (=6.5×102 counts/s percm2), which implied that a single photoelectron repre-sents 2.2×102 photons. This is consistent with theresults of the calculation. However, a difference mayexist due to overestimation of the light collecting effi-ciency of the lens system, without regard to the transmit-tance property and the variation in relative illumination,which depends on the image height. If these are consid-ered, Pmin when T=1 s can be estimated: 9.9×10−17

W/cm2 (=3.1×102 photon/s per cm2 at 630 nm).

3.2. Analysis for spatiotemporal characteristics

To evaluate the performance of the spatiotemporalcharacterization of ultraweak photon emission, the pro-cess of obtaining time–space correlation of photoelec-trons was examined. Optically attenuated light emittingdiodes (LEDs), which had an averaged intensity of 5.83counts/s, were used for the evaluation. Using a 100-Hzrectangular pulse, a pair of LEDs was directly operatedto blink alternately. Fig. 3(a) shows an image obtainedwith 30 min integration. The image was displayed withextraction of the area measuring 8.4×8.4 mm, and thediameter of an LED was 2 mm. In Fig. 3(b), the intensitytime courses of both LEDs were obtained from thesquare regions, as shown in Fig. 3(a), and the averagedintensities on the two LEDs were comparable. Thecorrelation function, calculated with a time resolution of100 ms from 56 k photoelectron data, which correspondsto the measuring time of 3.0×103 s, is shown in Fig. 3(c),in which the relevant phase delay between the two LEDs,corresponding to the operation frequency, is indicated.Fig. 3(d) is the same function calculated from 16 k data(4.4×102 s measuring time). Although the correlationfunction is relatively noisy, the phase delay is discernible.The minimum number of data necessary to obtaincorrelation depends on the dark count level. In this case,the dark count for the selected regions was 0.27 counts/s.

M. Kobayashi et al. / Journal of Neuroscience Methods 93 (1999) 163–168 167

3.3. Ultraweak photon emission images of a rat’s brainand its spatiotemporal characterization

A typical example of biophoton emission images of anormal rat brain, compared against the image obtainedafter cardiac arrest which was led by excess dose ofpentobarbital, and extracted time courses of photonemission, are shown in Fig. 4 with subtraction of thedark counts. The images displayed in Fig. 4(a,b) wereobtained with 1-h integration before and after cardiacarrest. The temporal variation of intensity of the wholearea of the brain is shown in Fig. 4(d) and time coursesof the four divided regions (Fig. 4c) in Fig. 4(e).

After cardiac arrest, the intensity of photon emissionwas reduced to approximately 60% of the normal con-dition, implying that the photon emission mechanism isassociated with cerebral blood flow.

Although temporal variations of intensity (Fig. 4e)indicate similar patterns throughout the area (Fig. 4d),the difference in intensity among the selected regions isrecognized. Localization of photon emission intensity isalso discernible from the image in Fig. 4(a), and thespatial distribution is speculated to represent the local-ization of ROS generation and could carry the informa-tion of the regional susceptibility to oxidative stress.

4. Discussion

The present study introduced a system for highlysensitive imaging and spatiotemporal analysis of ultra-weak photon emission and evaluated its performancethrough in vivo imaging of biophoton emission from arat’s brain. Spatiotemporal analysis of biophotons fromthe cortical surface indicates potential for in vivo ex-traction of physiological information from a brain.

The technique for spatiotemporal characterization ofultraweak photon emission in a two-dimensional pho-ton counting field may be applicable to many otheroptical measurements of brain function and physiology,not only for biophoton emission, but for example, forweak fluorescence or chemiluminescence measurementusing infused chemical makers. In particular, for micro-scopic measurements of dyed cells or transgenic cellsexpressing luciferase, a new technique for data acquisi-tion and processing that characterizes signal transduc-tion and cellular communication will be better suitedfor spatiotemporal correlation analysis of photo-electrons.

In vivo visualization of biophoton emission from thebrain has characteristics with respect to ROS monitor-ing and cellular injury. In general, brain tissue accountsfor approximately 20% of the whole body oxygen con-sumption and contains a comparatively high concentra-tion of polyunsaturated fatty acids. It is known to bevery susceptible to radical injuries subsequent to ROSgeneration. Oxidative stress, for example under postis-chemic reperfusion, leads to lipid peroxidation andoxidative injury (Traystman et al., 1991), focally affect-ing selectively vulnerable regions of the brain (Watsonet al., 1984; Bromont et al., 1988). Therefore in situidentification of the spatiotemporal properties of oxida-tive stress is most valuable and important for investiga-tions of radical injuries.

Ultraweak biophoton emission observed under nor-mal conditions suggests that intracellular respiration ofmitochondria for energy (ATP)-yielding metabolismparticipates in photon emission through leakage ofROS within the electron transport chain. It is suspectedthat an excited species for photon emission is formedthrough a radical reaction with ROS and intracellularsubstances. Particularly in the case of unsaturated fattyacids, excited species are generated as the result of alipid peroxidation process initiated by H2O2, metal-cat-alyzed OH�, or enzymatically produced ROS (Cadenaset al., 1984).

Although intrinsic photon emission without the infu-sion of chemical markers is extremely weak due to thelow efficiency of the excitation mechanism, the mea-surement of spontaneous photon emission has the ad-vantage of being a native reflection of physiological andpathological conditions. A newly developed two-dimen-sional photon counting imaging and analysis system of

Fig. 3. (a) An ultraweak photon emission image from two LEDsmeasured to evaluate system performance. Using a 100-Hz rectangu-lar pulse, two LEDs were switched to blink alternately. The imagesize is 8.4×8.4 mm, and the diameter of an LED is 2 mm. (b) Timecourses of emission intensity from the two LEDs, with an observationtime of 50 s, were extracted from the square regions indicated in (a).(c) Photoelectron correlation function extracted from two LED re-gions with a time resolution of 100 ms. This was calculated from adata number of 56 k, corresponding to the total measuring time of3.0×103 s. (d) Photoelectron correlation function obtained under thesame condition of (c) except that the total number of data points inthe calculation was from 16 k data (4.4×102 s).

M. Kobayashi et al. / Journal of Neuroscience Methods 93 (1999) 163–168168

Fig. 4. (a) Ultraweak photon emission image of a normal rat’s brain, compared with (b) an image obtained after cardiac arrest. The integrationtime is 1 h in both cases. The image size is 25×25 mm. (c) A photograph of the sample, indicating the measurement field. (d) Temporal changesof the intensity of ultraweak photon emission extracted from the whole region of the brain, indicated by the outer square in (c). (e) Temporalchanges of the intensity of ultraweak photon emission extracted from the four regions (c1–4) indicated in (c).

ultraweak photons has the potential to visualize thepathophysiological brain function in vivo.

References

Adamo AM, Llesuy SF, Pasquini JM, Boveris A. Brain chemilumines-cence and oxidative stress in hyperthyroid rats. Biochem J1989;263:273–7.

Amano T, Kobayashi M, Devaraj B, Usa M, Inaba H. Ultraweakbiophoton emission imaging of transplanted bladder cancer. UrolRes 1995;23:315–8.

Boveris A, Cadenas E, Reiter R, Filipkowski M, Nakase Y, ChanceB. Organ chemiluminescence: Noninvasive assay for oxidativeradical reactions. Proc Natl Acad Sci USA 1980;77:347–51.

Bromont C, Marie C, Bralet J. Increased lipid peroxidation invulnerable brain regions after transient forebrain ischemia in rats.Stroke 1988;20:918–24.

Cadenas E, Boveris A, Chance B. Low-level chemiluminescence ofbiological systems. In: Proyor WA, editor. Free Radicals in Biology,vol. 6. New York: Academic Press, 1984:211–42.

Cilent G. Photobiochemistry without light. Experientia 1988;44:572–6.Devaraj B, Usa M, Inaba H. Biophotons: ultraweak light emission from

living systems. Curr Opin Solid State Mater Sci 1997;2:188–93.Inaba H. Super-high sensitivity systems for detection and spectral

analysis of ultraweak photon emission from biological cells andtissues. Experientia 1988;44:550–9.

Kobayashi M, Devaraj B, Usa M, Tanno Y, Takeda M, Inaba H.Development and applications of new technology for two-dimen-sional space–time characterization and correlation analysis of

ultraweak biophoton information. Front Med Biol Eng1996;7:299–309.

Kobayashi M, Devaraj B, Usa M, Tanno Y, Takeda M, Inaba H.Two-dimensional imaging of ultraweak photon emission fromgerminating soybean seedlings with a highly sensitive CCD camera.Photochem Photobiol 1997a;65:535–7.

Kobayashi M, Takeda M, Yoshida M, Ito K-I, Usa M, Kato H, InabaH. The first demonstration of 2D-imaging of ultraweak biophotonemission from rat brain in vivo. Med Biol Eng Comput1997b;35(Suppl.):673.

Kobayashi M, Devaraj B, Inaba H. Observation of super-Poissonstatistics of bacterial (Photobacterium phosphoreum) biolumines-cence during the early stage of cell proliferation. Phys Rev E1998;57:2129–33.

Popp F-A. Biophoton emission. Experientia 1988;44:543–4.Popp F-A, Li KH, Mei WP, Galle M, Neurohr R. Physical aspects of

biophotons. Experientia 1988;44:576–85.Slawinski J. Luminescence research and its reaction to ultraweak cell

radiation. Experientia 1988;44:559–71.Traystman RJ, Kirsch JR, Koehler RC. Oxygen radical mechanisms

of brain injury following ischemia and reperfusion. J Appl Physiol1991;71:1185–95.

Usa M, Devaraj B, Kobayashi M, Takeda M, Ito H, Jin M, Inaba H.Detection and characterization of ultraweak biophotons from lifeprocesses. In: Ohzu H, Komatsu S, editors. Optical Methods inBiomedical and Environmental Sciences. Amsterdam: Elsevier,1994:3–6.

Van Wijk R, Schamhart DHJ. Regulatory aspects of low intensityphoton emission. Experientia 1988;44:586–93.

Watson BD, Busto R, Goldberg WJ, Santiso M, Yoshida S, GinsbergMD. Lipid peroxidation in vivo induced by reversible globalischemia in rat brain. J Neurochem 1984;42:268–74.