Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
BOLD fMRI Response of Early Visual Areas to Perceived Contrast in Human Amblyopia BRADLEY G. GOODYEAR,1,2 DAVID A. NICOLLE,3 G. KEITH HUMPHREY,4 AND RAVI S. MENON1,2,5 1 Laboratory for Functional Magnetic Resonance Research, The John P. Robarts Research Institute, London, Ontario N6A 5K8; 2Department of Medical Biophysics, The University of Western Ontario, London, Ontario N6A 5C1; 3Department of Ophthalmology, London Health Sciences Centre, London, Ontario N6A 5A5; 4Department of Psychology, The University of Western Ontario, London, Ontario N6A 5C2; and 5Department of Diagnostic Radiology and Nuclear Medicine, The University of Western Ontario, London, Ontario N6A 5A5, Canada Received 5 November 1999; accepted in final form 6 July 2000 Goodyear, Bradley G., David A. Nicolle, G. Keith Humphrey, and Ravi S. Menon. BOLD fMRI response of early visual areas to perceived contrast in human amblyopia. J Neurophysiol 84: 1907–1913, 2000. In this study, we used a temporal two-alternative forced choice psychophysical procedure to measure the observer’s perception of a 22% physical contrast grating for each eye as a function of spatial frequency in four subjects with unilateral amblyopia and in six subjects with normal vision. Contrast thresholds were also measured using a standard staircase method. Additionally, bloodoxygenation-level– dependent (BOLD) functional magnetic resonance imaging (fMRI) was used to measure the neuronal response within early visual cortical areas to monocular presentations of the same 22% physical contrast gratings as a function of spatial frequency. For all six subjects with normal vision and for three subjects with amblyopia, the psychophysically measured perception of 22% contrast as a function of spatial frequency was the same for both eyes. Threshold contrast, however, was elevated for the amblyopic eye for all subjects, as expected. The magnitude of the fMRI response to 22% physical contrast within “activated” voxels was the same for each eye as a function of spatial frequency, regardless of the presence of amblyopia. However, there were always fewer “activated” fMRI voxels during amblyopic stimulation than during normal eye stimulation. These results are consistent with the hypotheses that contrast thresholds are elevated in amblyopia because fewer neurons are responsive during amblyopic stimulation, and that the average firing rate of the responsive neurons, which reflects the perception of contrast, is unaffected in amblyopia. INTRODUCTION Amblyopia, clinically defined as a reduction in visual acuity in an otherwise healthy and properly corrected eye, is found in 1– 4% of the North American population (Ciuffreda et al. 1991; Daw 1995; Kiorpes and Movshon 1996; von Noorden 1990). The study of amblyopia in humans has almost entirely been by psychophysical means, where it has been demonstrated that the disorder produces a variety of spatial vision abnormalities (Carkeet et al. 1996; Hess et al. 1978; McKee et al. 1992; Reed et al. 1996; von Noorden 1990). A ubiquitous finding is a reduction in contrast sensitivity, the reciprocal of the contrast Address for reprint requests: B. G. Goodyear, Laboratory for Functional Magnetic Resonance Research, The John P. Robarts Research Institute, 100 Perth Dr., PO Box 5015, London, Ontario N6A 5K8, Canada (E-mail: [email protected]). www.jn.physiology.org required to visually detect a target (for review, see Ciuffreda et al. 1991), which for the normal eye is maximum between 2 and 5 cycles per degree (cpd) (see for example, Blakemore and Campbell 1969; Campbell and Robson 1968). For the amblyopic eye, the reduction in contrast sensitivity becomes more pronounced at higher spatial frequencies, and in some cases, maximum sensitivity occurs at a spatial frequency lower than that for the normal eye (Bradley and Freeman 1981; Hess and Pointer 1985; Hess et al. 1978). Contrast detection thresholds are dependent on the “noise” present in the visual system (e.g., Graham 1989). That is, neural firing rates (integrated over some region) must exceed some signal-to-noise ratio threshold in order for a contrast pattern to be detected by the observer. In the case of amblyopia, more contrast may be required to surpass this threshold signal-to-noise ratio. Decreases in signal could result from fewer responsive neurons during stimulation of the amblyopic eye or from neural firing rates that are less during amblyopic eye stimulation than during normal eye stimulation, while increases in noise could result from a loss of neural connections or a rearrangement of connections within the cortex (for discussion see Kiorpes and McKee 1999; Kiorpes et al. 1999; Levi 1991). In contrast-matching studies performed well above threshold, targets viewed separately with the amblyopic and normal eye appear to have the same contrast (Hess and Bradley 1980; Hess et al. 1983; Loshin and Levi 1983). A possible interpretation of this result is that once the detection signal-tonoise ratio of neural activity has been surpassed, the perception of contrast may depend on the average firing rate of the responsive neurons, which appears not be impaired in amblyopia (Levi 1991). Animal models have demonstrated that neural substrates of amblyopia are evident in visual areas as early as primary visual cortex (area V1) in the form of a reduction in binocular processing and/or a marked shift in the ocular dominance of neuronal response away from the amblyopic eye (Crawford and von Noorden 1979, 1980; Hubel et al. 1977; Kiorpes et al. 1998; LeVay et al. 1980; Movshon et al. 1987). Although studies of human amblyopia have also demonstrated a decrease The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 0022-3077/00 $5.00 Copyright © 2000 The American Physiological Society 1907 1908 GOODYEAR, NICOLLE, HUMPHREY, AND MENON FIG. 1. Method of determining the perception of 22% contrast as a function of spatial frequency. A: an example of 2 trials. The spatial frequency being tested (i.e., stimulus ‘1’) is presented for 500 ms, followed by a blank gray screen for 500 ms, and then the reference frequency (i.e., stimulus ‘2’), which is common to all trials is presented for 500 ms at 12, 17, 22, 27, or 32% contrast. The subject then chooses the target (‘1’ or ‘2’) that appeared to have the greater contrast. In the example 1st trial, the subject probably chooses ‘2’. In the next trial, the contrast of ‘1’ is kept at 22%, and the contrast of ‘2’ is randomly chosen from 1 of the above-mentioned 5. In this example 2nd trial, the subject probably chooses ‘1’ as having the greater contrast. B: an example of a response function corresponding to 1 test spatial frequency. The percentage of subject responses indicating the reference frequency (i.e., stimulus ‘2’) had the greater contrast is plotted vs. the actual contrast of stimulus ‘2’. The contrast of stimulus ‘2’ that matches the contrast of stimulus ‘1’ is taken as the 50% point on a curve fitted to the data. In this example, the 22% contrast test frequency appears to be the same as a 19.8% contrast reference frequency. One response function is obtained for each tested spatial frequency. in binocular interaction (Holopigian et al. 1986; Horton et al. 1997; Levi et al. 1979), histological studies of human visual cortex demonstrate no shift in ocular dominance away from the amblyopic eye (Horton and Hocking 1996; Horton and Stryker 1993). However, this may have been due to the onset of amblyopia being after the critical period of ocular dominance column development (Daw 1995). Neuroimaging studies of human amblyopia in living human brain using positron emission tomography (PET) (Demer et al. 1988) and single photon emission computed tomography (SPECT) (Kabasakal et al. 1995) have demonstrated that the measured signal within V1 during monocular stimulation of the amblyopic eye is less than that measured during monocular stimulation of the normal eye. However, no psychophysical assessment of the perception of contrast was determined. Magnetoencephalography (MEG) has demonstrated interocular differences in the magnitude of evoked magnetic responses to chromatic gratings throughout the entire occipital cortex of amblyopes; however, no correlation with contrast sensitivity of the amblyopic eye was found (Anderson et al. 1999). Because the magnitude of the measured signal within an image voxel during functional magnetic resonance imaging (fMRI) is thought to be correlated with the amount of neural activity within that voxel (for a review, see Ogawa et al. 1998), fMRI is well suited for noninvasively measuring the neuronal response to suprathreshold contrast. We investigated the perception of contrast using both psychophysical and fMRI measurements to understand the relationship between these very different methods and to determine the neural correlates of contrast perception in human amblyopia. METHODS Subjects Four subjects with unilateral amblyopia developed during early childhood or infancy were recruited through the Department of Ophthalmology at the London Health Sciences Center, London, Ontario, Canada. Each subject had been previously evaluated clinically by orthoptic assessment. During all experiments, each subject’s vision was optically corrected using their existing prescription lenses. Six subjects with no known visual deficits were recruited from the academic environment of the University of Western Ontario (UWO) to act as control subjects. All subjects had no previous experience with MR imaging. The experimental protocol was approved by the UWO Human Subjects Review Board. Visual stimuli Visual stimuli consisted of stationary vertical sinusoidal gratings at six spatial frequencies [0.5, 1, 2, 4, 8, 12 cycles per degree (cpd)] and at a mean luminance of 20 cd/m2. Contrast and luminance of each grating were measured using a Minolta CS-100 Chroma Meter (Minolta Camera, Osaka, Japan). Contrast was defined in the usual way as the maximum luminance of the grating minus the minimum luminance, divided by twice the mean luminance. Each grating was confined to a circle that subtended 12° of visual angle, and was reduced in contrast near the edge of the circle to eliminate sharp edges. The surrounding area of the display was gray at a luminance of 20 cd/m2. Visual stimuli were presented on a projection screen that was mounted onto the patient bed of the MR scanner, and was located 1.25 m from the subject’s eyes. An angled mirror, positioned above the subject’s eyes, provided a full view of the screen. Subjects wore a pair of liquid crystal shutter glasses (Milgram 1987), allowing the experimenter to control when each eye viewed the stimulus. Psychophysical measurements of the perception of contrast Psychophysical experiments were conducted while the subject lay in the MR scanner. The perception of 22% physical contrast was measured at each spatial frequency by obtaining the matching contrast of a standard reference frequency (1 cpd for amblyopes, 4 cpd for normals) using a temporal two-alternative forced choice procedure shown by example in Fig. 1A. The eye being tested was presented with one of the six test frequency gratings for 500 ms at 22% contrast with the eyepiece of the shutter glasses opaque for the fellow eye. This was followed by a 500-ms presentation of an isoluminant gray screen matched in mean luminance to the grating. Finally, the reference frequency, which did or did not differ in physical contrast from that of the test frequency, was presented for 500 ms. The subject was then asked to choose the grating that contained the greater contrast. This was repeated to obtain five responses at each of five contrasts of the FMRI RESPONSE TO CONTRAST IN HUMAN AMBLYOPIA reference frequency (12, 17, 22, 27, and 32%; this range of contrasts was selected on the basis of pilot studies), which were paired with the test frequency in random order. For one-half of the trials, the order of presentation was reversed. The perceived contrast of the test frequency grating was then obtained as the matching contrast of the reference frequency by taking the 50% point on a fitted curve to the corresponding response function (see Fig. 1B) (Wetherill and Levitt 1965). This procedure was then repeated for the remaining test frequencies, and then the whole procedure was repeated for the other eye. An additional measurement of the perception of contrast was made to compare contrast perception across eyes. To accomplish this, the 22% contrast reference frequency was presented to one of the eyes for 500 ms. The eyepiece of the glasses was then made opaque for that eye. Then, the fellow eyepiece was made transparent, and the reference frequency was shown to the other eye at one of the above five contrasts for 500 ms. The same forced-choice procedure was then used to match the contrast of the 22% reference frequency as seen with one eye to that seen with the other eye. Finally, contrast threshold measurements were taken for each eye separately using a standard staircase method. Because contrast perception measurements were obtained as matches to a fixed reference frequency, only the contrast threshold of the reference frequency was measured. fMRI All imaging experiments were performed on a Varian/Siemens Unity INOVA 4 Tesla whole-body MR scanner, equipped with 25 mT/m whole-body gradients. An 8-cm-diameter quadrature radio frequency (RF) surface coil was placed at the back of the subject’s head near the occipital pole to transmit and receive the MR signal. The subject’s head was immobilized using a well-padded head vice that was mounted onto the same platform that housed the RF coil. Seven imaging planes were prescribed parallel to the calcarine fissure of the visual cortex, which was identified within each subject with the aid of a sagittal localizer anatomical image. The functional image data acquisition scheme was a T*2-weighted, eight-segment, slice-interleaved, echo-planar imaging sequence (Menon and Goodyear 1999) [128 ⫻ 128 matrix, 14 ⫻ 14 cm field of view, 3 mm slice thickness, echo time (TE) ⫽ 15 ms, volume repetition time (TR) ⫽ 4 s, flip angle ⫽ 35°]. During image acquisition, each spatial frequency grating was presented separately and monocularly at 22% contrast, in random order, a total of three times, with each presentation lasting 3 s every 20 s. The remaining 17 s of each trial was a luminance-matched gray screen. At the end of the experiment, anatomical reference images were collected using a three-dimensional, T1-weighted FLASH imaging sequence [256 ⫻ 256 ⫻ 32 matrix, 14 ⫻ 14 ⫻ 4.8 cm field of view, inversion time (TI) ⫽ 0.5 s, TE ⫽ 6.3 ms, TR ⫽ 11.7 ms, flip angle ⫽ 11°]. Before data analysis, all images were motioncorrected using SPM96 (Friston et al. 1995), leaving five slices per subject for analysis. Including the setup time of the experiment, imaging time for anatomical localizer scans, anatomical reference scans, and a short break between psychophysical and imaging experiments, the total time to conduct all experiments for each subject was nearly 2 h. Hence, to relieve subject boredom and reduce subject motion, we used only five repetitions of the standard frequency within trials of the psychophysical tasks and three presentations of each spatial frequency target during imaging. Data analysis With the aid of the anatomical reference images, regions of interest (ROIs) for analysis were selected to include primary visual cortex (V1) and possibly other early visual areas (V2). All analyses were performed on an individual subject basis and separately for each eye. 1909 1. Classification of amblyopia for each subject by orthoptic assessment TABLE Refractive Correction Subject, Age (yr), Sex Strabismic Sphere Cylinder Axis, deg Snellen Deviation Eye (diopters) (diopters) (rotation) Acuity (Prism Diopters) Strabismus DA, 42, M MA, 56, F R L R L ⫹0.50 ⫹0.00 ⫹6.00 ⫹5.25 ⫹1.00 ⫹1.00 ⫹0.75 90 97 86 20/20 20/200 20/400 20/30 L.XT., 50 R.XT., 40 Strabismus/Anisometropia LL, 39, F CB, 51, F R L R L ⫹6.00 ⫹3.75 ⫹2.50 ⫹1.50 ⫹1.00 ⫹2.00 30 180 20/250 20/20 20/125 20/25 R.XT., 30 R.ET., 4 L.XT., left exotropia; R.XT., right exotropia; R.ET., right esotropia. Using a box-car input correlation function incorporating a temporal delay to allow for the intrinsic delay of the hemodynamic response, one functional map of image voxels showing a significant increase (P ⬍ 0.01) in MR signal above baseline was created for each and every spatial frequency. These maps were combined as a “logical OR” to create one map of voxels to be used in further analyses. The average fMRI response over the three presentations for each spatial frequency was then computed for each voxel (expressed as a percent increase in MR signal above baseline). To estimate the noise in the fMRI signal during visual stimulation, the magnitude of the MR signal as a function of time (i.e., the MR time course) for each image voxel within our selected ROIs (all voxels in the raw image data, not merely the resulting functional map) was Fourier transformed, multiplied by a high-pass frequency filter, and inverse-Fourier transformed to regain the MR time course. This method removed components of the MR signal that were in phase with the presentation of the visual stimuli (as well as all lower frequency signals), and maintained the high-frequency noise. The standard deviation of the magnitude of the MR signal for time points corresponding to stimulation of the eye was then computed as an estimate of the noise during stimulation of the eye. RESULTS The orthoptic assessment for each subject with amblyopia is summarized in Table 1. The severity of anisometropic and strabismic amblyopia varied across subjects, ranging from 4 to 50 prism diopters of strabismic deviation and a 0.25- to 2.25diopter difference in either spherical or cylindrical refractive correction. Only subjects with a spherical or cylindrical correction ⬎1 were considered anisometropic. Figure 2 shows each subject’s perception of 22% contrast as a function of spatial frequency measured with each eye for the four subjects with unilateral amblyopia as well as the average perception for the six subjects with normal vision. All subjects correctly perceived the reference frequency as a 22% contrast grating (i.e., physical contrast agreed with perceived contrast). One subject, subject CB, was unable to detect the 12 cpd target with the amblyopic eye. Thus data for subject CB at this spatial frequency was not analyzed. The curves in each graph (and in all subsequent similar graphs) were obtained by fitting the data, S, to a function of the form S ⫽ a b e ⫺c (1) 1910 GOODYEAR, NICOLLE, HUMPHREY, AND MENON FIG. 2. Psychophysical measurements of the perception of 22% contrast as a function of spatial frequency for 4 subjects with unilateral amblyopia and the average for 6 subjects with normal vision. Error bars represent SE. where is spatial frequency, and a, b, and c ⬎ 0 are fitted parameters. For all subjects with normal vision and for three of the four subjects with amblyopia, the perception of contrast was the same for both eyes. For each of these three amblyopes, there was little or no difference in the location of the maximum of the curve fitted to the data for each eye [DA: 2.5 cpd (preferred eye), 2.1 cpd (amblyopic eye); MA: 2.5, 1.9; CB: 1.8, 1.8]. Subject LL exhibited a decrease in the perception of contrast measured with the amblyopic eye, and this decrease was more pronounced at higher spatial frequencies. In this case, the maximum of the curve fitted to the amblyopic eye data occurred at 0.8 cpd, and at 2.4 cpd for the preferred eye data. Figure 3 shows that the magnitude of the fMRI response as a function of spatial frequency was the same for each eye for the subjects with normal vision and for the same three amblyopes that exhibited the same psychophysical response for each eye. However, the magnitude of the fMRI response to stimulation of the amblyopic eye for subject LL was smaller at higher spatial frequencies compared with the fMRI response to stimulation of the normal eye, in agreement with the psychophysical data. Figure 4 correlates the results of Figs. 2 and 3 to show that regardless of spatial frequency the magnitude of the fMRI response reflects the perception of contrast and is the same for each eye, even for subject LL. That is, even though the perception of 22% contrast measured with the amblyopic eye of LL was reduced, the correlation between perceived contrast and fMRI response was the same as that for the preferred eye. This remarkable relationship between fMRI response and level of contrast is consistent with a model that predicts a power relationship between the two at low contrast levels (Boynton et al. 1999), i.e., fMRI response ⬀ (perceived contrast)b, where b is usually ⬎2. For our data, b was 4.54 (R2 ⫽ 0.9). The threshold contrast of the reference frequency for the amblyopic eye was elevated for all subjects [DA: 2.4% (amblyopic eye), 1.5% (preferred eye); MA: 2%, 1.2%; CB: 1.9%, FIG. 3. Functional magnetic resonance imaging (fMRI) response to 22% physical contrast (expressed as a percentage increase in MR signal above baseline) of image voxels within early visual areas as a function of spatial frequency for 4 subjects with unilateral amblyopia and the average for 6 subjects with normal vision. Error bars represent SE. 1.5%; LL: 3%, 2.2%]. Figure 5A shows the perceived contrast as a function of spatial frequency expressed in multiple units of the threshold contrast for each eye (i.e., the perceived contrast divided by threshold contrast), averaged over the four subjects with amblyopia. Figure 5B shows the number of activated voxels that were in each of the functional maps created at each spatial frequency before they were combined. Although Fig. 4 suggests that the magnitude of the fMRI response reflects the perception of contrast and is not dependent on the presence of amblyopia, Fig. 5 suggests that the total neural activity within early visual areas contributing to a detectable fMRI response is not the same for each eye. The reduction in the number of activated voxels resulting from amblyopic eye stimulation was FIG. 4. The fMRI responses of Fig. 3 plotted against the corresponding psychophysical measurements of the perception of 22% contrast of Fig. 2 for the 4 subjects with unilateral amblyopia and the average for the 6 subjects with normal vision. Error bars represent SE. FMRI RESPONSE TO CONTRAST IN HUMAN AMBLYOPIA FIG. 5. A: average psychophysical measurements of the perception of 22% contrast expressed in multiples of threshold contrast for the 4 subjects with unilateral amblyopia. B: the average number of voxels within the fMRI maps as a function of spatial frequency for the 4 subjects with amblyopia (* P ⬍ 0.05). Error bars represent SE. significant (P ⬍ 0.05, paired t-test at each spatial frequency across subjects) only at higher spatial frequencies, but the trend is quite clear. In support of the finding that the total neural signal in early visual areas is smaller during stimulation of the amblyopic eye than it is during stimulation of the normal eye, it was found that on average across subjects with amblyopia, the underlying high-frequency noise in the fMRI signal during stimulation of the amblyopic eye was 0.20 ⫾ 0.04% (mean ⫾ SE). This did not differ from the noise during stimulation of the good eye (0.22 ⫾ 0.03%), nor different from the noise during stimulation of either eye of those subjects with normal vision (0.19 ⫾ 0.02%). DISCUSSION Using brief stimuli (⬍4 s) is key in high spatial resolution fRMI studies to maintain the spatial specificity of the fMRI response by preventing saturation of the local hemodynamic response (Das and Gilbert 1995; Malonek and Grinvald 1996) and minimizing adaptation effects. We have previously shown that an unsaturated hemodynamic response is also more likely to maintain proportionality between neural activity and bloodoxygenation-level– dependent (BOLD) fMRI signal (Menon and Goodyear 1999). Our findings may not have been possible using a sustained visual stimulus, as any linearity between neural activity and fMRI response would have likely been lost. As well, relatively minor changes in the perceived contrast of the stimulus would have been masked by the spreading of the vascular response to areas unassociated with the neural activity. Previous fMRI analysis techniques for the study of contrast response include preselecting voxels for analysis on the basis of functional localizer experiments using high stimulus contrasts (Boynton et al. 1999). This removes any statistical biases toward any one contrast level by fully saturating the fMRI response. In our study, however, only 22% physical contrast was used. We did not use the reference frequency to preselect voxels for analysis because this would have favored that spatial frequency. Our approach of creating a statistical map for each 1911 and every spatial frequency and then combining the maps to create one group of voxels for further analysis removes any bias toward one spatial frequency. Of the four subjects shown in Table 1, subject LL demonstrated the most severe case of anisometropia (i.e., the greatest difference between the eyes in terms of refractive correction). The reduction in contrast perception for subject LL (Fig. 2) as measured with the amblyopic eye is consistent with previous findings (Hess and Pointer 1985) demonstrating that in moderate-to-severe cases of anisometropia, there is an elevation in the level of contrast at which the performance of the two eyes equalizes. Hence, our visual stimulus parameters may have been such that a 22% contrast was not sufficient for subject LL to overcome the perceptual deficits of the ambylopic eye at low contrast. This was shown as well in the fMRI response to stimulation of the amblyopic eye with the same visual targets (Fig. 3). These data, as well as the data for the other three amblyopes and for the subjects with normal vision, clearly demonstrate that fMRI response, and consequently the level of neural activity, in early visual areas of the visual cortex correlates with the perception of contrast. These results are consistent with previous work showing that the fMRI response reflects the level of contrast (Goodyear and Menon 1998; Tootell et al. 1998), and further that the coding of contrast within monocular neurons is spared in amblyopia (Kiorpes et al. 1998; Movshon et al. 1987). Of course, the voxels of our MR images contain many types of neurons, not merely monocular neurons. Hence we cannot claim that the measured fMRI response is a direct measure of the activity of neurons that respond preferentially to stimulation of one eye. The exact contribution of the activity of these types of neurons to the fMRI signal is difficult to determine given the practical spatial resolution restrictions of fMRI. Nonetheless, our results show that magnitude of neural activity per image voxel is not impaired in the presence of amblyopia when contrast is sufficient to equate the perception of that contrast across the eyes. Figure 5B illustrates, however, that there is a reduction in the number of “activated” image voxels in the presence of amblyopia. Because no change in image signal-to-noise is expected across the eyes during the experiment, the presence of fewer voxels in our functional maps for stimulation of the amblyopic eye as a function of spatial frequency cannot be explained as an imaging artifact. Statistical methods of generating functional maps depend on signal magnitude and signal variance to determine whether a voxel is significantly “active.” Our analysis has shown that the underlying high-frequency noise in the fMRI signal during visual stimulation is unchanged in the presence of amblyopia, and thus a difference in signal magnitude must be responsible for a difference in the number of activated voxels. A reduction in the number of voxels in our functional maps for amblyopic eye stimulation is consistent with low spatial resolution results obtained from PET (Demer et al. 1988), SPECT (Kabasakal et al. 1995), and MEG studies (Anderson et al. 1999), since total (or pooled) activity measured by these methods in the visual cortex would be related to the product of the magnitude of our fMRI signal change and the number of voxels in our maps. A reduction in pooled neural activity within the visual cortex has also been proposed as a possible mechanism for elevated contrast thresholds in amblyopia (for a review, see Levi 1991). When perceived contrast is expressed in multiple units of the threshold contrast (Fig. 5A), 1912 GOODYEAR, NICOLLE, HUMPHREY, AND MENON our data seem to suggest that the number of detectable voxels at some statistical threshold using fMRI (Fig. 5B) may be related to the difference in threshold contrast between the normal and amblyopic eyes since that difference is signal-tonoise dependent. This is merely suggestive, of course, since our measurements are made at 22% contrast, not at threshold. However, we speculate that at all perceptually equivalent contrasts, the relative number of detectable voxels during stimulation of the normal and amblyopic should remain approximately constant. Once the perception of contrast becomes reduced for the amblyopic eye, the relative number of image voxels will probably favor normal eye stimulation since we expect that the magnitude of the fMRI signal will decrease for the amblyopic eye (as for example, subject LL in Fig. 3) as contrasts approach threshold. The direct investigation of contrast threshold response versus suprathreshold contrast response using fMRI will require stimuli restricted to the fovea or to a narrow range of eccentricities since unlike psychophysical judgements of suprathreshold contrast (see for example, Cannon and Fullenkamp 1988), contrast threshold judgements are heavily eccentricity dependent (e.g., Robson and Graham 1981; Savoy and McCann 1975). Further elucidation of the neural substrates governing contrast sensitivity loss in amblyopes must be performed with detailed retinotopy (Tootell et al. 1998) to examine the contribution of areas outside V1. In addition, fMRI at higher spatial resolution will allow an evaluation of the response of the ocular dominance columns (Menon and Goodyear 1999) to stimulation of amblyopic eye. Nonetheless, our current results clearly demonstrate that fMRI of the visual cortex of amblyopes is sensitive to the reduction in pooled neural activity in response to stimulation of the amblyopic eye as a function of spatial frequency. As well, fMRI is also sensitive to neural populations whose contrast coding seems to be spared at suprathreshold contrast, even in the presence of amblyopia. Although, based on the results of this study, we cannot specify the neural mechanisms underlying amblyopia, we have demonstrated that fMRI is a useful noninvasive tool in the investigation of amblyopia at the cortical level. The authors thank Dr. David Heeger at Stanford University for input and helpful suggestions. This work was supported by National Eye Institute Grant 1R01EY-11551 and the Medical Research Council (Canada). REFERENCES ANDERSON SJ, HOLLIDAY IE, AND HARDING GF. Assessment of cortical dysfunction in human strabismic amblyopia using magnetoencephalography (MEG). Vision Res 39: 1723–1738, 1999. BLAKEMORE C AND CAMPBELL FW. On the existence of neurones in the human visual system selectively sensitive to the orientation and size of retinal images. J Physiol (Lond) 203: 237–260, 1969. BOYNTON GM, DEMB JB, GLOVER GH, AND HEEGER DJ. Neuronal basis of contrast discrimination. Vision Res 39: 257–269, 1999. BRADLEY A AND FREEMAN RD. Contrast sensitivity in anisometropic amblyopia. Invest Ophthamol Vis Sci 21: 467– 476, 1981. CAMPBELL FW AND ROBSON JG. Application of Fourier analysis to the visibility of gratings. J Physiol (Lond) 197: 551–556, 1968. CANNON MW JR AND FULLENKAMP SC. Perceived contrast and stimulus size: experiment and simulation. Vision Res 28: 695–709, 1988. CARKEET A, LEVI DM, AND MANNY RE. Visibility of motion in infant vernier displays, using adult subjects. Ophthalmic Physiol Opt 16: 239 –242, 1996. CIUFFREDA KJ, LEVI DM, AND SELENOW A. Amblyopia: Basic and Clinical Aspects. Boston, MA: Butterworth-Heinemann, 1991. CRAWFORD MLJ AND VON NOORDEN GK. The effects of short-term experimental strabismus on the visual system in Macaca mulatta. Invest Ophthalmol Vis Sci 18: 496 –505, 1979. CRAWFORD MLJ AND VON NOORDEN GK. Optically induced concommitant strabismus in monkeys. Invest Ophthalmol Vis Sci 19: 1105–1109, 1980. DAS A AND GILBERT CD. Long-range horizontal connections and their role in cortical reorganization revealed by optical recording of cat visual cortex. Science 375: 780 –784, 1995. DAW NW. Visual Development. New York: Plenum, 1995. DEMER JL, VON NOORDEN GK, VOLKOW ND, AND GOULD KL. Imaging of cerebral blood flow and metabolism in amblyopia by positron emission tomography. Am J Ophthalmol 105: 337–347, 1988. FRISTON KJ, JEZZARD P, AND TURNER R. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 2: 189 –210, 1995. GOODYEAR BG AND MENON RS. Effect of luminance contrast on BOLD fMRI response in human primary visual areas. J Neurophysiol 79: 2204 –2207, 1998. GRAHAM NVS. Visual Pattern Analyzers. New York: Oxford Univ. Press, 1989. HESS RF AND BRADLEY A. Contrast perception above threshold is only minimally impaired in human amblyopia. Nature 287: 463– 464, 1980. HESS RF, BRADLEY A, AND PIOTROWSKI L. Contrast coding in amblyopia. I. Differences in the neural basis of amblyopia. Proc R Soc Lond B Biol Sci 217: 309 –330, 1983. HESS RF, CAMPBELL FW, AND GREENLAUGH T. On the nature of the neural abnormality in human amblyopia: neural aberrations and neural sensitivity loss. Pflügers Arch 337: 201–207, 1978. HESS RF AND POINTER JS. Differences in the neural basis of human amblyopia: the distribution of the anomaly across the visual field. Vision Res 25: 1577–1594, 1985. HOLOPIGIAN K, BLAKE R, AND GREENWALD MJ. Selective losses in binocular vision in anisometropic amblyopes. Vision Res 26: 621– 630, 1986. HORTON JC AND HOCKING DR. Pattern of ocular dominance columns in human striate cortex in strabismic amblyopia. Vis Neurosci 13: 787–795, 1996. HORTON JC, HOCKING DR, AND KIORPES L. Pattern of ocular dominance columns and cytochrome oxidase activity in a macaque monkey with naturally occurring anisometropic amblyopia. Vis Neurosci 14: 681– 689, 1997. HORTON JC AND STRYKER MP. Amblyopia induced by anisometropia without shrinkage of ocular dominance columns in human striate cortex. Proc Natl Acad Sci USA 90: 5494 –5498, 1993. HUBEL DH, WIESEL TN, AND LEVAY S. Plasticity of ocular dominance columns in monkey striate cortex. Philos Trans R Soc Lond B Biol Sci 278: 377– 409, 1977. KABASAKAL L, DEVRANOGLU K, ARSLAN O, ERDIL TY, SONMEZOGLU K, USLU I, TOLUNM H, ISITMAN AT, OZKER K, AND ONSEL C. Brain SPECT evaluation of the visual cortex in amblyopia. J Nucl Med 36: 1170 –1174, 1995. KIORPES L, KIPER DC, O’KEEFE LP, CAVANAUGH JR, AND MOVSHON JA. Neuronal correlates of amblyopia in the visual cortex of macaque monkeys with experimental strabismus and anisometropia. J Neurosci 18: 6411– 6424, 1998. KIORPES L AND MCKEE SP. Neural mechanisms underlying amblyopia. Curr Opin Neurobiol 9: 480 – 486, 1999. KIORPES L AND MOVSHON JA. Amblyopia: a developmental disorder of the central visual pathways. Cold Spring Harb Symp Quant Biol 61: 39 – 48, 1996. KIORPES L, TANG C, AND MOVSHON JA. Factors limiting contrast sensitivity in experimentally amblyopic macaque monkeys. Vision Res 39: 4152– 4160, 1999. LEVAY S, WIESEL TN, AND HUBEL DH. The development of ocular dominance columns in normal and visually deprived monkeys. J Comp Neurol 191: 1–51, 1980. LEVI DM. Spatial vision in amblyopia. In: Spatial Vision, edited by Regan D. London, UK: MacMillan, 1991, p. 212–238. LEVI DM, HARWERTH RS, AND SMITH EL. Humans deprived of normal binocular vision have binocular interactions tuned to size and spatial frequency. Science 206: 852– 854, 1979. LOSHIN DS AND LEVI DM. Suprathreshold contrast perception in functional amblyopia. Doc Ophthalmol 55: 213–236, 1983. MALONEK D AND GRINVALD A. Interactions between electrical activity and cortical microcirculation revealed by imaging spectroscopy: implications for functional brain mapping. Science 272: 551–554, 1996. FMRI RESPONSE TO CONTRAST IN HUMAN AMBLYOPIA MCKEE SP, SCHOR CM, STEINMAN SB, WILSON N, KOCH GG, DAVIS SM, HSU-WINGES C, DAY SH, CHAN CL, AND MOVSHON JA. The classification of amblyopia on the basis of visual and oculomotor performance. Trans Am Ophthalmol Soc 90: 123–144, 1992. MENON RS AND GOODYEAR BG. Submillimeter functional localization in human striate cortex using BOLD contrast at 4 Tesla: implications for the vascular point-spread function. Magn Reson Med 41: 230 –235, 1999. MILGRAM P. A spectacle-mounted liquid-crystal tachistoscope. Behav Res Methods Inst Comp 19: 449 – 456, 1987. MOVSHON JA, EGGERS HM, GIZZI MS, HENDRICKSON A, KIORPES L, AND BOOTHE RG. Effects of early unilateral blur on the macaque’s visual system. III. Physiological observations. J Neurosci 7: 1340 –1351, 1987. OGAWA S, MENON RS, KIM S-G, AND UGURBIL K. On the characteristics of functional magnetic resonance imaging of the brain. Annu Rev Biophys Biomol Struct 27: 447– 474, 1998. 1913 REED MJ, STEEVES JK, STEINBACH MJ, KRAFT S, AND GALLIE B. Contrast letter thresholds in the non-affected eye of strabismic and unilateral enucleated subjects. Vision Res 36: 3011–3018, 1996. ROBSON JG AND GRAHAM N. Probability summation and regional variation in contrast sensitivity across the visual field. Vision Res 21: 409 – 418, 1981. SAVOY RL AND MCCANN JJ. Visibility of low-spatial-frequency sinewave targets: dependence on number of cycles. J Opt Soc Am 65: 343–350, 1975. TOOTELL RB, HADJIKHANI NK, VANDUFFEL W, LIU AK, MENDOLA JD, SERENO MI, AND DALE AM. Functional analysis of primary visual cortex (V1) in humans. Proc Natl Acad Sci USA 95: 811– 817, 1998. VON NOORDEN GK. Binocular Vision and Ocular Motility. St. Louis, MO: Mosby, 1990. WETHERILL GB AND LEVITT H. Sequential estimation of points on a psychometric function. Br J Math Stat Psychol 18: 1–10, 1965.