The Evolution of Optical Signals in Human and Rodent Cortex

3, 202–208 (1996) 0022

NEUROIMAGE ARTICLE NO.

The Evolution of Optical Signals in Human and Rodent Cortex ANDREW F. CANNESTRA, ANNE J. BLOOD, KEITH L. BLACK,*

AND

ARTHUR W. TOGA

Laboratory of Neuro Imaging, Division of Brain Mapping, Department of Neurology, and *Division of Neurosurgery, Department of Surgery, University of California at Los Angeles School of Medicine, 710 Westwood Plaza, Los Angeles, California 90095-1769 Received December 5, 1995

The time course of optical intrinsic signals was examined in order to characterize the evolution of response in human and rodent cortex. Both subtraction/ratio and principal component analyses were used to construct time-course curves. The time course began at a prestimulus baseline, responded with a finite delay, overcompensated, reduced to a maintenance level, and then disappeared. The magnitude, spatial involvement, and principal components demonstrated similar time-course curves both in human and in rodent. For acute stimuli, peak response was reached between 2 and 3 s and returned to baseline by 6 s poststimulation. The shape of the time-course curve is consistent with the need to satisfy neuronal demand and the contributions of vascular smooth muscle properties to the response behavior. The temporal delays and nonlinear phenomena observed in the time-course curves are consistent with a hydraulic model of neurovascular supply/demand behavior. r 1996 Academic Press, Inc.

INTRODUCTION Several indicators of changes in perfusion have been used to localize functional activity in neural tissue. Investigators have demonstrated changes in regional cerebral blood flow (rCBF) in response to somatosensory stimulation (Fox et al., 1988), motor tasks (Grafton et al., 1991), and mental activities (Petersen et al., 1988). Although these measurements describe the site of activation, available techniques have been unable to fully characterize the development, maintenance, and decay of this response due to limitations in resolution. As a result, temporal and spatial behaviors of blood delivery to active cortex are not well understood. Optical imaging of intrinsic signals (OIS) is a technique that measures changes in cortical reflected light due, in part, to vascular effects (Grinvald et al., 1986). Its high spatial (µm) and temporal (ms) resolution lends itself to improved observation of activity-dependent vascular responses. OIS temporal/spatial data obtained from postactivation studies thus allows the 1053-8119/96 $18.00 Copyright r 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

investigation of interactions and capacities of neurovascular responses (Frostig et al., 1990; Blood et al., 1995). A number of techniques have been used to describe the behavior of neurovascular physiology. Hyder et al. (1995) and Kwong et al. (1992) reported functional magnetic resonance imaging (fMRI) responses to peripheral stimulation of somatosensory cortex with peak responses near 3 s in rodent and 5 s in human. Hurtig et al. (1994) provided PET numerical simulations also suggesting human rCBF changes on the order of seconds poststimulation. Sitzer et al. (1994) measured human rCBF and flow-velocity changes using transcranial Doppler ultrasonography (TCD) with maximal poststimulation response very similar to rodent fMRI. In addition, Villringer et al. (1993) measured hemodynamic changes in human cortex (near 8 s poststimulation) in response to peripheral stimulation using near infrared spectroscopy (NIRS). Cox et al. (1993) used vascular dyes and latex beads in the rodent to examine the cortical vascular response appearing within seconds after peripheral somatosensory stimulation. Similarly, in rodent cortex, Narayan et al. (1995) provided evidence for a vascular contribution to the optical intrinsic signal using dyes showing a 2- to 3-s time course following peripheral somatosensory stimulation. Shevelev et al. (1993) and Santori et al. (1992) used thermoencephaloscopy (TES), which measures minute localized temperature changes as an indication of metabolic and vascular responses, in rodents and found peak responses varying between 2 and 10 s depending upon peripheral stimulation. Although the timings measured with these techniques are consistent, their resolution limitations make time-course studies difficult. As a result, the evolution of poststimulation response has not been fully described. fMRI and PET are limited by acquisition time and spatial resolution. TCD is spatially limited due to vascular anatomy (macroscopic vessels only) while NIRS and TES are spatially and temporally limited due to instrumentation and the requirement for thermal conduction through bone. In contrast, OIS provides considerable acquisition flexibility resulting in millisecond and micrometer resolutions.

202

EVOLUTION OF OPTICAL SIGNALS IN CORTEX

Optical intrinsic signals were first demonstrated in vitro by Hill and Keynes (1949) and have been subsequently used as a functional neuroimaging technique in rodent (Blood et al., 1995), cat (Ts’o et al., 1990), monkey (Frostig et al., 1990), and human (Haglund et al., 1992; Toga et al., 1995a,b; Cannestra et al., 1995). Optical signals in animals and humans colocalize with electrically active cortex (Frostig et al., 1990; Narayan et al., 1994a,b; Toga et al., 1995a,b), as well as with cytochrome oxidase-stained barrels in rodent (Blood et al., 1995). The etiology of the intrinsic signal includes reflectance changes from several optically active processes (Cohen, 1973), rather than indicating neuronal firing directly. Haglund et al. (1994) suggested that vascular dynamics and cellular swelling are both components of the reflectance change. Narayan et al. (1995) provided evidence that intrinsic signals are closely related to increases in cerebral blood volume. It is generally accepted that the observed contribution of each component to the optical signal may be, in part, wavelength dependent (Frostig et al., 1990; Malonek and Grinvald, 1995). Optical imaging has been used to help understand cortical physiology in the visual (Frostig et al., 1990), somatosensory (Grinvald et al., 1986), and language systems (Haglund et al., 1992). We used OIS imaging to reveal perfusion and other related metabolic responses of human and rodent sensorimotor cortex to peripheral somesthetic stimulation (Blood et al., 1995; Toga et al., 1995b). In this study, we developed time-course curves of magnitude, spatial involvement, and the principal components of optical intrinsic signals in order to characterize the evolution of optical response in human and rodent cortex. We discuss findings comparing different modalities, species, and paradigms. MATERIALS AND METHODS Similar methods were used in all human and rodent studies. Details are provided in other publications (Toga et al., 1995b; Blood et al., 1995). Briefly, we exposed the cortical field and imaged (200-ms exposure) with a slow scan CCD camera (Series 200, Photometrics, Tucson, AZ or Model TE/CCD-576EFT, Princeton Instruments, Trenton, NJ) mounted over suitable optics. Voltage-stabilized white light was used to epiilluminate the cortex. Images were acquired through transmission filters at 610 (600–620) nm for human studies and 850 (840–860) nm (Corion Corp., Holliston, MA) for rodent. Data acquisition (baseline and poststimulation images) was synchronized with monitored electrocardiographic and pneumographic waveforms. Human Studies We measured optical reflectance changes over sensorimotor cortex in seven anesthetized patients undergo-

203

ing surgical resection of parietal tumors. The imaging camera, illumination, and microscope were positioned and stabilized after the cortex was exposed. Following craniotomy and reflection of the dura, images were collected through the optics of the Zeiss operating microscope in response to transcutaneous electrical stimulation of the median or ulnar nerves (2 s, 15.5– 17.0 mA) using a bipolar electrode and Grass Instruments (S-12; Quincy, MA) stimulator. Trials contained eight runs and included an equal number of nonstimulated interleaved controls. Rodent Studies Seventeen adult male Sprague–Dawley rats (Charles River Laboratories) weighing between 200 and 350 g were used for the rodent experiments. Rats were placed in a stereotaxic frame, their heads were shaved, and their skulls were cleaned. A small metal scraping instrument (Biomedical Research Instruments, Inc., Rockville, MD) was then used to uniformly thin the bone over the primary somatosensory region of the right hemisphere. Silicone oil was then applied in order to make the bone more translucent. A single whisker, C1, was stimulated via a 12-V electromechanical motorized nudger with an angle of deflection approximately 30°, velocity 1.2 m per second anterior to posterior (frequency 10 Hz, deflections lasted 70 ms; acute stimulations: duration 0.3 s, i.e., three deflections; sustained stimulation: duration 10 s, i.e., 100 deflections). Statistical Image Analysis Subtraction and ratio image analysis (Narayan et al., 1994a) was performed to demonstrate magnitude and spatial involvement time courses. Calculation procedures are outlined below. Principal component analysis (PCA) was applied to both the subtraction and the ratio-analyzed data and the originally acquired data sets. Since PCA represents a combined spatial and magnitude time course, we present principal components as a measure of these time courses. Signal magnitude was defined as the average pixel intensity in a statistically defined region of interest (ROI). ROIs were first created from the averaged ratio images at each timepoint by performing a 3-pixel Gaussian blur, equalizing and thresholding at the mean pixel value 1 1 STD. This ROI was then superimposed upon the original averaged ratio image and the magnitude was calculated (Blood et al., 1995). The timepoint with the largest calculated magnitude was considered to be the maximal optical response for the trial. This ‘‘maximal’’ ROI was then superimposed upon the other images of the trial in order to calculate the respective magnitude at each timepoint. In one human subject the signal to noise ratio (SNR) was very low. In this case, we calculated the differences between tempo-

204

CANNESTRA ET AL.

ral images to remove noise artifacts. The magnitude (and spatial involvement described below) was then calculated as defined above. The spatial involvement was defined as the pixel count in a second statistically defined ROI. Similar to the ROI determined for the magnitude calculation, the spatial ROI was calculated from the average ratio images by performing a 3-pixel Gaussian blur and thresholding. Images in an experimental trial were thresholded at the level determined for the individual maximum response image within that trial (as determined by the magnitude calculation above, mean pixel value 11 STD). Principal Component Analysis Determining the eigenvalues and eigenvectors. The mathematical procedure for PCA has been outlined in detail in previous reports (Geladi et al., 1989; Yap et al., 1994). Briefly, given a set of images [Xt50 to N ] (image size 192 3 144) from timepoints 0 to N, we construct a single matrix [XS] (size 27648 3 11; 192 3 144 5 27648, 0 to 10 5 11 timepoints) in which the columns represent a pixel intensity of a specific image and the rows represent the intensity time course of a single pixel from the image set. Matrix [XS]T is transposed, and the covariance matrix [C] (11 3 11) determined from the relation below. [C] 5 [Xs][Xs]T The eigenvalues (li, where i 5 1 to 11) and eigenvectors (Vli) were determined by the Jacobi transformations technique. Choosing the significant eigenvectors. Once the Vli’s were determined, the significant Vli’s were chosen dependent upon the SNR of the images. When the SNR was high, the li’s with the single largest order of magnitude were used to select the Vli’s. When the SNR was low (one human case), we normalized all of the Vli’s and calculated the resultant squared difference from the established time-course curve (from subtraction and ratio statistical analysis). This calculation yielded a correlation constant, the lowest of which indicated the best fit to the established time-course curve and therefore the appropriate Vli. Reconstruction of the image data set. The matrix [XS] was then reconstructed by summing the contributions of each significant Vli at each pixel, n

[XR] 5

o ([X ] · V s

li)Vli,

i5m

where m and n are the actual numbers of the significant eigenvectors (i.e., for a reconstruction using Vl2 and Vl3, m 5 2 and n 5 3). [XR] was then decomposed into

the original data set by separating the columns of matrix [XR] into individual 192 3 144 images. RESULTS AND DISCUSSION We found the magnitude, spatial involvement, and principal component time-course curves calculated from subtraction and ratio analysis to be very similar between and within a species. As seen in Fig. 1, commencement of signal occurs by 1 s, peaks between 2 and 3 s, and decays by 6 s. The temporal resolution of human data is limited by synchronization with electrocardiographic waveforms (approximately 1/s). However, the human temporal response displays the same characteristics observed in the rodent optical response curve acquired at 700-ms resolution. The differences in slopes between the spatial extent and magnitude curves are 1.6% to peak and 3.7% for decay in humans, and in rodent are 11.9% to peak and 10.5% for decay. The interspecies differences are 11.1 and 2.8% to peak and 16.2 and 23.4% for decay, magnitude and spatial extent, respectively. The interspecies differences in the principal component curves are 7.4% to peak and 11.1% for decay. The Shape of the Time-Course Curves The time course began at a prestimulus baseline, responded with a finite delay, overcompensated (initially overresponding above subsequent steady-state levels), reduced to a maintenance level, and then disappeared (Figs. 1, 2, and 3).1 No oscillatory behavior was observed. These approximate 0.75- to 1.50-s delay and overcompensation phenomena also have been observed with other techniques (Ngai et al., 1988, 1995; Sitzer et al., 1994). Observation of the time-course curves was enhanced by the use of PCA, allowing rapid, nonbiased mathematical analysis and noise reduction. The principal component is an appropriate representa-

1 Figure 3 is an animation sequence of optical response data in both human and rodent. It may be accessed via the internet through the UCLA Laboratory of Neuro Imaging home page (http://www.loni. ucla.edu) under the subheading of Optical Imaging. The figure illustrates the spatial and temporal relationships observed in Figs. 1 and 2 during the evolution and decay of the optical response. Animations were constructed by a linear interpolation between individual images. (A) The human acute stimulation optical response was constructed from five trials of 14 frames to acute median nerve stimulation (2-s transcutaneous stimulation, pulse interval 0.2 s, pulse duration 200 ms/pulse, 15.5- to 17-mA current). Glare artifacts over the craniotomy margin were removed. (B) The rodent acute stimulation optical response consisted of nine trials of 9 images, C1 whisker stimulation by a motorized nudger (0.3 s at 10 Hz, 3 deflections). (C) The rodent long duration stimulation optical response consisted of six trials of 19 images, C1 whisker stimulation by a motorized nudger (10 s at 10 Hz, 100 deflections). All three animations may be played simultaneously in order to compare the similarities between species and paradigms.

EVOLUTION OF OPTICAL SIGNALS IN CORTEX

205

ouar et al., 1995), and OIS (Carmona et al., 1995; Kaplan et al., 1995; Hanson et al., 1995) functional imaging techniques. The factor analysis approach of PCA increased sensitivity and reduced noise (near a factor of 3). The correlation of the eigenvectors to actual time-course curves provided an objective method of successful signal and noise separation. We observed (Fig. 2) an initial overcompensation of vascular response leading to the maintenance of a lower response state during prolonged stimulation. The curve indicates that initial blood flow response to active cortex is three to four times the required maintenance level of the tissue (7 to 10 s). The reduction from peak to the maintenance level begins with a rapid decay phase (3.00 to 7.50 s) whose rate is faster than the subsequent slow decay phase by close to a factor of 3 (7.50 to 10.50 s). Ngai et al. (1988) has also observed the slow decay to baseline after cessation of prolonged stimulus (10 s), as well as a decay delay (approximately 0.50 to 1.25 s) between stimulus termination (10 s) and the initiation of slow decay (11.25 s). Previous studies from our laboratory have suggested a minimum stimulation threshold level (two whisker deflections) in order to obtain a vascular response (Blood et al., 1995). Differences between Species

FIG. 1. Comparison of the average magnitude, spatial extent, and principal component time courses from optical response data in human and rodent, respectively. Optical responses were recorded after peripheral stimulation (human: 2-s transcutaneous median or ulnar nerve stimulation, pulse interval 0.2 s, pulse duration 200 ms/pulse, 15.5- to 17-mA current; rodent: 0.3-s electromechanical motorized nudger, C1 whisker, 10 Hz, deflections lasted 70 ms, i.e., three deflections). Commencement of signals occurred by 1 s, peaked between 2 and 3 s, and decayed by 6 s. (A) The slopes to peak for the human time-course curves (0.00 to 2.00 s) were 47.4, 45.8, and 40.0%/s for magnitude, spatial involvement, and principal component, respectively. The slopes of decay (2.00 to 5.00 s) were 219.4, 223.1, and 219.4%/s for magnitude, spatial involvement, and principal component, respectively. (B) The slopes to peak for the rodent time-course curves (0.38 to 1.75 s) were 36.3, 48.2, and 32.7%/s for magnitude, spatial involvement, and principal component, respectively. The slopes of decay (2.45 to 4.55 s) were 235.6, 246.07, and 230.5%/s for magnitude, spatial involvement, and principal component, respectively. Error bars indicate standard errors.

tion of temporal functions since PCA is not biased by the determination of spatial constraints (such as ROI). It has previously been applied to the analysis of PET (Boulanouar et al., 1995; Pederson et al., 1994; Yap et al., 1994), SPECT (Houston et al., 1994), fMRI (Boulan-

Differences between species time-course curves may be attributed either to the 610- to 850-nm filtering difference between experiments or to the craniotomy/ thin bone preparation difference between experiments. The 610/850 difference is potentially significant since 610-nm filtration allows greater influence of the hemoglobin (Hb)-dependent vascular response than 850 nm (Frostig et al., 1990). Filtration of 850-nm reflected light reduces the contribution of Hb, presumably emphasizing more scattering from cellular swelling, blood vessel dilation, and increased intravascular volume (Frostig et al., 1990; Narayan et al., 1994a; Blood et al., 1995). Accordingly, when the human and rodent principal component curves are superimposed, the human 3and 4-s points are not within their respective standard errors to the rodent curve. This probably results from increased venous signal detection of Hb at 610-nm observation. The craniotomy/thin bone preparation difference is also important since, in contrast to the thin bone preparation, the craniotomy affords direct observation of cortex and increased sensitivity. On the other hand the craniotomy preparation adds significant cortical movement which can decrease the SNR. Narayan et al. (1994a) reported consistent timecourse optical response to forelimb electroshock and whisker deflection. Similarly, Blood et al. (1995) described similar optical responses for stimulations terminating before peak response. Above differences in duration and nature of stimulation between species may therefore minimally contribute to observed time-course differences.

206

CANNESTRA ET AL.

FIG. 2. The time-course evolution during a sustained whisker stimulus (10 s electromechanical motorized nudger, C1 whisker, 10 Hz, deflections lasted 70 ms, i.e., 100 deflections) in a single rodent. Stimulus onset occurred at 0 s. The images within the trial were reconstructed using the first principal component. The principal component is graphed in relation to the stimulus duration (percentage change versus time in seconds). The timecourse began at a prestimulus baseline (0.00 s), responded with a finite delay (0.75–1.50 s), overcompensated (peak at 3.00 s), reduced to a maintenance level (near 8 s), and then disappeared (by 12.75 s, 2.75 s poststimulus termination). The slopes were 43.5, 215.5, 24.3, and 26.9%/s for rise to peak (0.75 to 3.00 s), fast (2.63 to 7.13 s) and slow (7.13 to 10.13 s) reduction to maintenance level, and decay to baseline (10.50 to 12.75 s), respectively. The lower right indicates approximate cortical representations as determined by barrel and forelimb optical responses (representations adopted from Chapin and Lin, 1984). Color bar indicates reflectance decrease 3 1024. Scale bar, 1 mm.

The human OIS time-course curve is consistent with the temporal response observed from human TCD (Sitzer et al., 1994) as well as rodent OIS and the vascular literature (Ngai et al., 1988, 1995). However, both TCD and human OIS time courses present a 2- to 6-s more rapid evolution than human fMRI studies (Kwong et al., 1992; Bandettini et al., 1993; Frahm et

al., 1992). In contrast to these visual and motor system activation studies, human OIS responses were observed to activate small regions of somatosensory cortex (0.5 to 1 cm2). These regions are near or under single voxel fMRI resolution, potentially explaining this discrepancy. Time courses obtained from fMRI, OIS, and TCD are not necessarily in disagreement, but

207

EVOLUTION OF OPTICAL SIGNALS IN CORTEX

may rather represent differing components of neurovascular response. Prolonged stimulations in humans were not performed due to intraoperative constraints. However, the human/rodent acute stimulation similarities were in agreement with TCD (Sitzer et al., 1994). Proposed Hydraulic Model Overview. The response curves reported here are consistent with a theoretical hydraulic model where neurovascular behavior may be limited by fluid dynamics and physical constraints such as vessel conformation, proximity, dilatation capacities, signal delays, and sensitivity to feedback mechanisms. Assuming the vasculature is stable morphologically (macroscopically), dynamic changes may only occur locally. Furthermore, the dilatation capacity of the vessels may be limited both spatially and temporally. Spatially, the vessels are observed to only dilate within a specific range (Ngai et al., 1988) presumably determined by anatomical constraints. Temporally, the vascular smooth muscle relaxation delay (Ignarro et al., 1981) may play a critical role in the time course of the response. Model implications. The morphologic and spatial limitations may induce vascular involvement larger than active cortical areas due to the regional involvement of upstream parent vessels (Ngai et al., 1988), possible recruitment of multiple vascular tributaries (Cox et al., 1993), and the subsequent development of complex flow patterns. Narayan et al. (1995) illustrated with dyes that vascular fluorescence exceeds the area involved in neuronal stimulation. Temporal delays may be responsible for time courses on the order of seconds, as well as nonlinearities (responses not proportional to input, e.g., overcompensations) of the system. The vasculature may have evolved to compensate for this time constraint by oversupplying the neurons initially, causing the observed overcompensation/luxury perfusion behavior (Figs. 1 and 2). Simultaneous stimulations may induce a complex response dictated by the temporal and spatial relationships between active cortex. Coexisting active areas of cortex may compete for the vascular response, resulting in the division of resources (blood pressure and flow) and complex nonlinear behavior. We reported previously the response to simultaneous stimulation of neighboring cortex is not simply additive, rather, it is nonlinear (Toga et al., 1995a,b; Cannestra et al., 1995). Blood and Toga (1995) also suggested, through competition experiments, the existence of more widespread nonlinear optical behavior. This nonlinearity is potentially a common response in physiologically interacting cortex and requires further study. The observed behavior is consistent with a need to satisfy neuronal demand for vascular supply. Threshold, delay, overcompensation, and slow decay phenomena are compatible with the development of a metabolic substance and its diffusion to trigger a vascular re-

sponse (Moncada et al., 1991). Similarities between observed time courses and that of vascular smooth muscle relaxation (Ignarro et al., 1981) are consistent, possibly explaining OIS response delays. The equivalence of the interspecies time-course curves suggests that differences in vascular anatomy, absolute diameter, and absolute vascular distance from active cortex may not be major factors in determining response. Similarly, within the sensorimotor system, differences in cortical anatomy, underlying cytoarchitecture, and absolute cortical representation may not exert measurable effects upon the time course of neurovascular response. Conclusion We have described the magnitude, spatial extent, and principal component time courses in order to characterize the optical intrinsic signal evolution in human and rodent cortex. Following a prestimulus baseline, the system responded with a finite delay, overcompensated, reduced to a maintenance level, and then disappeared. The shape of the time-course curve is consistent with the need to satisfy neuronal demand and the contributions of vascular smooth muscle properties to the response behavior. The temporal delays and nonlinear phenomena observed in the time-course curves are consistent with a hydraulic model of neurovascular supply/demand behavior. ACKNOWLEDGMENTS The authors thank John S. Burton and Drs. Cloughesy and Rubinstein for their assistance. Andrew F. Cannestra is supported, in part, by a Medical Scientist Training Program grant (GM08042). This work is supported by the National Institute for Mental Health (MH/NS52083) and the National Sciences Foundation (BIR9322434). The authors thank Andrew Lee for his image to bit stream (animation) assistance.

REFERENCES Bandettini, P. A., Jesmanowicz, A., Wong, E. C., and Hyde, J. S. 1993. Processing strategies for time-course data sets in functional MRI of the human brain. Magn. Reson. Med. 30:161–173. Boulanouar, K., Henry, B., Demonet, J. F., and Berry, I. 1995. Principal component analysis on PET and MRI images. J. Cereb. Blood Flow Metab. 15(Suppl. 1):S602. Blood, A. J., and Toga, A. W. 1995. Modulation of optical signals by introduction of a competing stimulus. Neurosci. Abstr. 21:1900. Blood, A. J., Narayan, S. M., and Toga, A. W. 1995. Stimulus parameters influence characteristics of optical intrinsic responses in somatosensory cortex. J. Cereb. Blood Flow Metab. 15(6):1109– 1121. Carmona, R. A., Hwang, W. L., and Frostig, R. D. 1995. Wavelet analysis for brain function imaging. IEEE Trans. Med. Imaging 14(3):556–564. Cannestra, A. F., Toga, A. W., and Black, K. L. 1995. The temporal/ spatial evolution of optical signals in humans. Neurosci. Abstr. 21:1208. Chapin, J. K., and Lin, C.-S. 1984. Mapping the body representation

208

CANNESTRA ET AL.

in the SI cortex of anesthetized and awake rats. J. Comp. Neurol. 229:199–213. Cohen, L. B. 1973. Changes in neuron structure during action potential propagation and synaptic transmission. Physiol. Rev. 53(2):373–418. Cox, S. B., Woolsey, T. A., and Rovainen, C. M. 1993. Localized dynamic changes in cortical blood flow with whisker stimulation corresponds to matched vascular and neuronal architecture of rat barrels. J. Cereb. Blood Flow Metab. 13(6):899–913. Fox, P. T., Raichle, M. E., Mintun, M. A., and Dence, C. 1988. Non-oxidative glucose consumption during focal physiologic neural activity. Science 241:462–464. Frahm, J., Bruhn, H., Merboldt, K.-D., and Hanicke, W. 1992. Dynamic MR imaging of human brain oxygenation during rest and photic stimulation. J. Magn. Reson. Imaging 5:501–505. Frostig, R. D., Lieke, E. E., Ts’o, D. Y., and Grinvald, A. 1990. Cortical functional architecture and local coupling between neuronal activity and the microcirculation revealed by in vivo high-resolution optical imaging of intrinsic signals. Proc. Natl. Acad. Sci. USA 87:6082–6086. Geladi, P., Isaksson, H., Lindqvist, L., Wold, S., and Esbensen, K. 1989. Principal component analysis of multivariate images. Chem. Intell. Lab. Syst. 5:209–220. Grafton, S. T., Woods, R. P., Mazziotta, J. C., and Phelps, M. E. 1991. Somatotopic mapping of the primary motor cortex in humans: Activation studies with cerebral blood flow and positron emission tomography. J. Neurophysiol. 66(3):735–743. Grinvald, A., Lieke, E., Frostig, R. D., Gilbert, C. D., and Wiesel, T. N. 1986. Functional architecture of cortex revealed by optical imaging of intrinsic signals. Nature 324:361–364. Haglund, M. M., Hochman, D. W., Meno, J. R., Ngai, A. G., and Win, H. R. 1994. Mechanisms underlying the intrinsic signal during optical imaging of rat somatosensory cortex. Neurosci. Abstr. 20:1264. Haglund, M. M., Ojemann, G. A., and Hochman, D. W. 1992. Optical imaging of epileptiform and functional activity in human cerebral cortex. Nature 358:668–671. Hanson, C., Fu, Q.-G., Chen, G., and Ebner, T. J. 1995. Functional organization of the rat cerebellar cortex elucidated by principal component analysis of optical signals. Neurosci. Abstr. 21:1910. Houston, A. S., Kemp, P. M., and Macleod, M. A. 1994. A method for assessing the significance of abnormalities in HMPAO brain SPECT images. J. Nucl. Med. 35:239–244. Hill, D. K., and Keynes, R. D. 1949. Opacity changes in stimulated nerve. J. Physiol. (London) 108:278–281. Hurtig, R. R., Hichwa, R. D., O’Leary, D. S., Boles Ponto, L. L., Narayana, S., Watkins, G. L., and Andreasen, N. C. 1994. Effects of timing and duration of cognitive activation in [15O]water PET studies. J. Cereb. Blood Flow Metab. 14(3):423–430. Hyder, F., Kaiser, M. G., Behar, K. L., Blamire, A. M., Chase, J. R., Martin, M. A., Rothman, D. L., During, M. J., and Shulman, R. G. 1995. Time-courses of rapid functional magnetic resonance imaging and cerebral blood flow during fore-paw stimulation in rat. J. Cereb. Blood Flow Metab. 15(Suppl. 1):S489. Ignarro, L. J., Lippton, H., Edwards, J. C., Baricos, W. H., Hyman, A. L., Kadowitz, P. J., and Gruetter, C. A. 1981. Mechanism of vascular smooth muscle relaxation by organic nitrates, nitrites, nitroprusside and nitric oxide: Evidence for the involvement of S-nitrosothiols as active intermediates. J. Pharmacol. Exp. Ther. 218:739–749. Kaplan, E., O’Brain, E. V., Orbach, D., Everson, R., and Gordon, J. 1995. Time course and wavelength dependence of the intrinsic

optical signal from the mammalian visual cortex investigated with principal component analysis. Neurosci. Abstr. 21:393. Kwong, K. K., Belliveau, J. W., Chesler, D. A., Goldberg, I. E., Weisskopf, R. M., Poncelet, B. P., Kennedy, D. N., Hoppel, B. E., Cohen, M. S., Turner, R., Cheng, H.-M., Brady, T. J., and Rosen, B. R. 1992. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc. Natl. Acad. Sci. USA 89:5675–5679. Malonek, D., and Grinvald, A. 1995. The spatial and temporal relationship between cortical electrical activity and responses of the microcirculation during sensory stimulation. Neurosci. Abstr. 21:772. Moncada, S., Palmer, R. M., and Higgs, E. A. 1991. Nitric oxide: Physiology, pathophysiology, and pharmacology. Pharmacol. Rev. 43:109–142. Narayan, S. M., Esfahani, P., Blood, A. J., Sikkens, L., and Toga, A. W. 1995. Functional increases in cerebral blood volume over somatosensory cortex. J. Cereb. Blood Flow Metab. 15:754–765. Narayan, S. M., Santori, E. M., Blood, A. J., Burton, J. S., and Toga, A. W. 1994a. Imaging optical reflectance in rodent barrel and forelimb sensory cortex. NeuroImage 1:181–190. Narayan, S. M., Santori, E. M., and Toga, A. W. 1994b. Mapping functional activity in rodent cortex using optical intrinsic signals. Cereb. Cortex 4:195–205. Ngai, A. C., Ko, K. R., Morii, S., and Winn, H. R. 1988. Effect of sciatic nerve stimulation on pial arterioles in rats. Am. J. Physiol. 254(1):H133–H139. Ngai, A. C., Meno, J. R., and Winn, H. R. 1995. Simultaneous measurements of pial arteriolar diameter and laser–Doppler flow during somatosensory stimulation. J. Cereb. Blood. Flow. Metab. 15:124–127. Pedersen, F., Bergstorm, M., Bengtsson, E., and Langstrom, B. 1994. Principal component analysis of dynamic positron emission tomography images. Eur. J. Nucl. Med. 21:1285–1292. Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., and Raichle, M. E. 1988. Positron emission tomographic studies of the cortical anatomy of single-word processing. Nature 331:585–589. Santori, E. M., Narayan, S., Gorbach, A. M., and Toga, A. W. 1992. Thermal mapping of rat cortical function. Neurosci. Abstr. 18:864. Shevelev, I. A., Tsicalov, E. N., Gorbach, A. M., Budko, K. P., and Shaev, G. A. 1993. Thermoimaging of the brain. J. Neurosci. Methods 46:49–57. Sitzer, M., Knorr, U., and Seitz, R. 1994. Cerebral hemodynamics during sensorimotor activation in humans. J. Appl. Physiol. 77(6): 2804–2811. Toga, A. W., and Cannestra, A. F. 1995a. The temporal/spatial evolution of optical signals in human cortex. Hum. Brain Map. Suppl. 1:170. Toga, A. W., Cannestra, A. F., and Black, K. L. 1995b. The temporal/ spatial evolution of optical signals in human cortex. Cereb. Cortex 5(6):561–565. Ts’o, D. Y., Frostig, R. D., Lieke, E. E., and Grinvald, A. 1990. Functional organization of primate visual cortex revealed by high resolution optical imaging. Science 249:417–420. Villringer, A., Planck, J., Hock, C., Schleinkofer, L., and Dirnagl, U. 1993. Near infrared spectroscopy (NIRS): A tool to study hemodynamic changes during activation of brain function in human adults. Neurosci. Lett. 154(1–2):101–104. Yap, J. T., Chen, C.-T., Cooper, M., and Treffert, J. D. 1994. Knowledgebased factor analysis of multidimensional nuclear medicine image sequences. Proc. SPIE 2168:289–297.