|Home | About | Journals | Submit | Contact Us | Français|
We characterized the perceptual, functional, and structural abnormalities associated with retinal ischemia during a cotton wool spot episode and its sequelae. The border of the visually salient field anomaly mirrored the quantitatively measured relative scotoma. Results of resolution perimetry and high resolution imaging indicated that there was a substantial loss of retinal ganglion cells within the affected region. A disruption in retinal nerve fiber arrangement was found at the cotton wool spot and within the arcuate relative scotoma. The presence of the arcuate relative scotoma is consistent with the hypothesis of failed signal transmission along the axons that pass through the cotton wool spot. The different levels of loss associated with the arcuate and focal scotomas indicate different underlying pathologies.
Cotton wool spots (CWS) are associated with a wide range of systemic diseases such as diabetes and hypertension (Brown, Brown, Hiller, Fischer, Benson & Magargal, 1985; Schmidt, 2008). An isolated CWS is believed to be caused by an acute focal occlusion of a single retinal arteriole. The reduced perfusion in the inner retina causes ischemia of the nerve fiber layer which in turn disrupts axoplasmic flow. As a result axoplasmic debris accumulates in the retinal ganglion cell (RGC) axons, and this is thought to interfere with signal conduction (McLeod, 2005). This build up of debris is visible ophthalmoscopically as a reflective “cotton wool spot” (McLeod, Marshall, Kohner & Bird, 1977). Although most CWSs are asymptomatic (Schmidt, 2008), one report has described localized scotomas at the site of CWS (Bek & Lund-Andersen, 1991), while others report arcuate scotomas consistent with signal transmission failures of the RGC axons that pass through that CWS lesion (Alencar, Medeiros & Weinreb, 2007; Shami & Uy, 1996). These two types of visual defect may reflect different consequences of the local perfusion failure. The focal scotoma conjugate with the CWS presumably reflects hypoxia in the inner retinal neurons under the CWS. The arcuate scotoma likely reflects a failure of signal transmission in the axon bundles that pass through the ischemic region of the retinal nerve fiber layer (RNFL). Color vision testing suggests that RGCs carrying chromatic information may be affected more than cells carrying achromatic information (King-Smith, Vingrys & Benes, 1987).
Since little is known about the functional and structural characteristics of such defects, or their ability to spontaneously resolve, the case reported here provided an opportunity to examine both the visual disturbances associated with CWSs and the various hypotheses for the pathophysiology of CWS (McLeod, 2005). A vision scientist with extensive experience at performing peripheral visual tasks observed a sudden onset of a unilateral visual anomaly in the near peripheral field. The subject immediately documented its visual characteristics and we followed those observations with more quantitative perimetry, clinical evaluation, and retinal imaging. The ability to document the full chronology of this pathology in an observer experienced in performing peripheral vision tasks, in combination with access to a psychophysical and retinal imaging laboratories, made this a unique opportunity to describe the visual changes that accompanied the emergence and recovery of visual function and retinal structure associated with a CWS.
A 52 year old male patient noticed a visual disturbance in the inferior temporal field of his right eye in March, 1999. At the time he was self-reported to be in good general health. Manual perimetry performed by the patient revealed an arcuate relative scotoma. No clinical examination was performed on that occasion and the scotoma resolved. On 11-Jun-04, a similar visual disturbance was observed in a different visual field location in the left eye. Clinical evaluation by a retinal specialist at that time revealed no obvious anomaly. Another field anomaly was noticed on 18-Sept-04 at a similar location as the March 1999 event. Manual perimetry was performed the next day. A fundus examination was performed by a retinal specialist on 27-Sept-04 who diagnosed a single CWS in the retinal location indicated by the manual perimetry. Medical history was unremarkable without any systemic diseases and additional blood tests were negative. Visual acuity was 20/20 in each eye with spectacle correction. Fundus photography showed a focal retinal lesion (CWS) located 10° from the fovea in the superior nasal retina (Fig. 1a). There was no optic disk abnormality. Fluorescein angiography of the right eye manifested focal hypofluorescence (Fig. 1b) consistent with either occlusion of a retinal arteriole or simply masking of retinal fluorescence by the absorbent materials accumulated within in the CWS (McLeod, 2005). Funduscopically all other visible retinal blood vessels were normal. Automated visual field tests (Humphrey 24-2) of the right eye were unremarkable.
One month later, fundus photography of the same region revealed shrinkage of the CWS (Fig. 1c). Similar to previous clinical studies, StratusOCT (Carl Zeiss Meditec, Dublin, CA) images revealed a hyper-reflective tissue in the axon layer coincident with the lesion site (Fig. 1d) (Kozak, Bartsch, Cheng & Freeman, 2006; Kozak, Bartsch, Cheng & Freeman, 2007). At 8 months follow up there was resolution of the CWS (Fig. 1e). Specialized psychophysical measurements were repeated during the ensuing 7 months. Long term structural sequalae were measured using high-resolution retinal imaging four years after the initial event.
Three different psychophysical approaches were used to evaluate the shape, severity and perceptual manifestations of the visual field anomaly.
The subject recorded the appearance of the arcuate anomaly while changing fixation from light to dark fields and while blinking. Also, the margin of the highly visible field anomaly was mapped on a sheet of graph paper that was placed 57cm from the eye so that 1 cm on the paper corresponded to 1° of visual angle. While fixating a point on the paper, a small black dot was swept across the visual field to locate the border of the visual field anomaly. Using the same apparatus, the blind spot was also mapped. The appearance of objects within the field anomaly was documented.
Computer-based kinetic perimetry of the lesion border was performed by presenting a low contrast (5%) grey circular disk (0.3° diameter) viewed against a white background (luminance = 107 cd/m2) on a computer screen. Targets with contrast of 10% or more were unsuitable for mapping the arcuate relative scotoma because they were clearly visible throughout most of the visual field anomaly. The focal absolute scotoma at the CWS was mapped by a high contrast (88%) circular target (0.3° diameter). Disk position was controlled in 0.1° steps. While fixating a mark at the center of the monitor, the subject used keyboard commands to perform a method-of-limits paradigm in which the stimulus is moved across the lesion from seen to unseen, and then returning from unseen to seen. The lesion edge was indicated by the midpoint of the two locations where the spot appeared to cross the border. Mapping was performed three weeks after the field anomaly was first noticed, 2 months later, and four years after the first event.
Spatial resolution deficits were evaluated using resolution perimetry, a technique that reveals the sampling density of RGCs in the peripheral retina (Anderson, Wilkinson & Thibos, 1992; Thibos, 1998). The stimulus was a circular patch of sinusoidal grating (mean luminance = 67 cd/m2 ; contrast = 97%) displayed at one of seven possible test locations either within the lesion (L) or at adjacent regions of normal visual field (N) (Fig. 2). At a viewing distance of 4.75m, the stimulus subtended 0.6° at locations N1, N2, L1, and L2, and 1.1° at N3, N4, and L3. A control experiment was performed with the stimulus presented at the corresponding areas of L2, L3, and N3 in the fellow eye. The maximum resolvable spatial frequency (cycles/degree) was determined by the method of limits using a spatial aliasing criterion (Thibos, Walsh & Cheney, 1987b). The subject’s task was to increase the spatial frequency of a grating until it began to produce perceptual aliasing. Then the reverse procedure was conducted in which the spatial frequency of an aliased pattern was reduced until it appeared veridical. A total of five reversal points were recorded and the mean of these five reversal points was taken as the estimate of sampling-limited visual resolution. The experiment was performed three weeks after the lesion onset and again 6 months later.
Long term structural deficits were measured in 2008 using an AOSLO (Burns, Tumbar, Elsner, Ferguson & Hammer, 2007). Confocal retinal images were obtained using a super luminescent diode (SLD) light source with a 50nm bandwidth centered at 840nm. Confocal apertures were 12µm and 24µm relative to the retina for cone photoreceptor imaging and RNFL imaging, respectively. A dental impression was used to stabilize the subject’s head movements and the pupil was dilated with 0.5% tropicamide. A set of steering mirrors were used to position the beam sequentially across the CWS region, using the 2004 fundus images as a guide. Imaging of a comparable area in the corresponding healthy retinal region in the same eye served as an internal control. Informed consent was obtained after a full explanation of the procedures and its risks. This study protocol was approved by Indiana University Institutional Review Board and complied with the requirements of the Declaration of Helsinki.
Frames from the resulting high resolution images were averaged offline. Multiple small field images were then stitched together to create a retinal montage using Adobe Photoshop CS3 (Adobe Systems Inc, San Jose, CA). Cone photoreceptor positions within the montage were determined semi-automatically using a custom Matlab program (MathWorks, Natick, MA) (Chui, Song & Burns, 2008a). We estimated the cone Nyquist frequency assuming an hexagonal array and a retinal magnification factor of 0.290 mm/deg (Hirsch & Miller, 1987; Snyder & Miller, 1977).
Infrared SLO fundus imaging and SDOCT imaging were also performed to obtain cross-sectional measurements of the retina (Spectralis HRA+OCT, Heidelberg Engineering, Heidelberg, Germany). RNFL thickness was estimated from the images across the CWS region and in comparable regions in the unaffected regions of both eyes. The axial and lateral resolutions of the SDOCT were approximately 7µm and 14µm respectively. For this experiment, each b-scan covering 30° of the central retina was composed of 1,536 equally spaced a-scans. RNFL thickness was measured by delineating the first signal from the inner limiting membrane and the signal from the anterior boundary of the RGC layer.
On 18-Sept-04, the subject experienced a visual disturbance that was similar perceptually to an afterimage of a bright light source. As with afterimages (Carpenter, 1972; Darwin, 1818), the contrast polarity depended on the background illumination, e.g. when looking at the bright sky, the disturbance appeared darker than the background but when looking at the dark earth, the disturbance looked brighter than the background. Unlike ordinary after-images, the visualization lasted continuously for several weeks.
The visual disturbance followed an arcuate path from the horizontal midline in the nasal field, sweeping inferiorly to a point just short of the blind spot (Fig. 3). The width of the arcuate disturbance was approximately 1°–2°, and within this narrow region light sensitivity was abnormally low but not absent and thus we use the term “relative scotoma”. Objects within the relative scotoma appeared mottled and shimmered similar to spatial aliasing as reported previously by this same observer (Thibos et al., 1987b). Suprathreshold targets within the relative scotoma appeared fragmented as if obscured by a perforated screen, similar to the undersampled spatial aliases seen in peripheral vision when viewing sinusoidal gratings too fine to be resolved (Thibos, 1998). As the circular disk passed across the relative scotoma, it looked like the full moon, partially obscured momentarily by broken clouds. This suggested that some, but not all, retinal neurons remained functional in the scotomatous area.
Near the far temporal end of the arcuate relative scotoma, a small, absolute scotoma was located about 3° short of the blind spot. When combined with the subsequently captured fundus image (Fig. 3), it can be seen that the focal absolute scotoma coincides with the location of the CWS, but exhibited a different shape. The shape and size of the arcuate and focal scotomas were reconfirmed several times over the following 4 years from Oct-04 to Oct-08.
After a few weeks, the subject was no longer aware of the arcuate relative scotoma or the small, localized absolute scotoma where the CWS occurred. However, the arcuate relative scotoma could be visualized as a dull, gray arc by blinking repeatedly. Blinking emphasizes the dark phase since the lids are open more than closed. Blinking in the dark did not produce a subjective visualization, indicating the effect was not an artifact of lid motion. Four years after the event, the arcuate region still appeared dark against a bright background and bright against a dark background. However, these were transient sensations that faded after a few seconds and were not obvious in daily life. Spatial distortion of text was noticeable when attending critically to the affected area of the visual field. The arcuate scotoma could be visualized by blinking throughout the four years following the initial event.
Low-contrast computer-based kinetic perimetry revealed an arcuate relative scotoma (Fig. 4), that corresponded closely with the location of the visual disturbance map shown in Fig. 3. The kinetic perimetry maps generated in October and December of 2004 were almost identical over most of the scotoma, with evidence of a slight contraction in the December map near to the CWS (Fig. 4). In addition to the large arcuate relative scotoma detectible with a low contrast target, a small focal absolute scotoma was measured coincident with the CWS using a high contrast target (>90%). A small region (represented by the 4-sided polygon in Fig. 4) was found to be completely blind on 3-Dec-04. This focal scotoma was also mapped using a hand held high contrast spot on 16-Oct-08 (gray ellipse in Fig. 4). On this date, the scotoma had an elliptical shape (approximately 1.5° diameter) with two tiny sub-regions of vision near its center (white circles in Fig. 4).
Resolution acuity for a sampling-limited task is shown in Table 1 for each of the 7 testing areas measured on two sessions separated by 6 months. At the time of the first test, 16 days after the visual disturbance was first noticed, resolution acuity at test location L1 was 3-fold lower than that at control location N1. RGC density and cone density should be the same at these two locations (Curcio & Allen, 1990; Curcio, Sloan, Kalina & Hendrickson, 1990). Acuity at location L1 was low relative to control area N2 located at the same eccentricity but further from the visual streak where cell density should be lower. Similarly acuity at L2 was lower than at N3, despite the fact that L2 is closer to the fovea and would normally have higher cell density and higher acuity. Six months later, resolution acuity had improved but not to normal levels. Perceptually, a well-resolved grating seen veridically at a greater eccentricity was clearly seen non-veridically when moved to the abnormal region of the visual field reinforcing the notion that this region of the retina was being undersampled.
High resolution AOSLO images of cone photoreceptors and RNFL in the vicinity of the former CWS and a corresponding healthy retinal region are shown in Fig. 5. Both the affected and control regions had cone photoreceptor density of ~7,000 cells/mm2 at 10° retinal eccentricity, in agreement with previous results from normal eyes (Chui, Song & Burns, 2008b; Curcio et al., 1990). In contrast, the AOSLO images of the RNFL in the affected eye showed an absence of the RNFL striation typically seen in normal eyes (Hogan, Alvardo & Weddell, 1971), with the RNFL appearing dark and blurred and lacking radiating RNFL bundles which are visible just above and below the defect. Cone photoreceptor density near the visual streak at test locations N1 and L1 (see Fig. 2) were ~20,000 cells/mm2 and ~18,000 cells/mm2 respectively. Fig. 5c and 5f provides possible evidence of a terminated arteriole (identified in Fig. 5f with arrow) consistent with the region of hypofluorescence seen in Fig. 1.
Infrared SLO fundus imaging and SDOCT imaging were performed on the affected and fellow eye in Sept-08. There was no abnormality of the optic disc or fovea in either eye. However there was an RNFL bundle defect in the affected eye starting from the CWS region and ending at the horizontal raphe of the retina (Fig. 6). This RNFL bundle defect appeared as a dark band. Fig. 7 and and88 show the SDOCT images obtained at the corresponding locations in the affected and in the fellow eye. Fig. 7c shows that the RNFL was less distinct and less reflective at the CWS region, RNFL thickness was 60% thinner at the CWS than in the corresponding location of the fellow eye. In addition, the RGC layer in the affected eye is barely noticeable at the site of the CWS. Although AOSLO images showed no photoreceptor mosaic abnormality in the CWS, SDOCT images revealed a focal thickening of the outer nuclear layer below the CWS in the affected eye as compared to the corresponding region on the fellow eye. Similarly, vertically oriented SDOCT b-scans through the arcuate lesion in the affected eye and fellow eye reveal RNFL thinning (Fig. 8).
Although the CWS and region of hypofluorescence (Fig. 1) were small (1° by 1.5°) and localized in the superior nasal retina, the patient observed an extensive arcuate visual disturbance (Fig. 3) with a shape that matched the distribution of RGCs with axons passing through the local region of the CWS (Hogan et al., 1971). Of particular note is the abrupt termination of this disturbance along the nasal horizontal midline where RGC fibers form the raphe (Hogan et al., 1971). The shape of this disturbance, therefore, is consistent with the hypothesis of failed signal transmission along the axons that pass through the CWS region. In addition to this arcuate band, there was a small absolute scotoma located in the superior temporal field that mapped directly onto the CWS and the region of possible hypoperfusion (McLeod, 2005) (Fig. 1 and and3).3). With the exception of a pair of very small points of vision (0.25° diameter), the subject was completely blind in this region, and showed no sign of recovery over the next 4 years. The different shapes and levels of severity of the arcuate and focal scotomas indicate different underlying pathologies.
CWSs consist of visible accumulation of axonal cytoplasmic debris secondary to a failure of both orthograde and retrograde axonal transport (McLeod et al., 1977) caused by a local perfusion failure in the RNFL (Dollery, Henkind, Paterson, Ramalho & Hill, 1966). The perfusion failure is generally thought to follow an occlusion of a retinal arteriole (Dollery et al., 1966; McLeod et al., 1977) that would normally perfuse a small region of the RNFL and the underlying inner retina. It is likely, therefore, that the focal absolute scotoma represents a loss of function in the inner retinal neurons at the CWS site due to ischemia (Dollery et al., 1966; McLeod et al., 1977). The fact that some vision remains within the arcuate relative scotoma indicates that some of the axons passing through the CWS remain active. These axons may survive the local RNFL perfusion failure because they acquire oxygen from the vitreous (McLeod et al., 1977). It is also possible that the physiology of some types of axons are resistant to a very localized perfusion failure. Another possibility is that axon paths to the optic disk vary slightly and thus some axons from the affected arcuate region might have gone around the ischemic region at the CWS.
The visual disturbance within the arcuate relative scotoma reported by our subject mirror those seen when an afterimage is generated (Carpenter, 1972; Darwin, 1818), and when a central absolute scotoma is present (Burke, 1999; Craik, 1966) and therefore might have a similar explanation. In all three cases, the region with reduced sensitivity was perceived as dark when viewed against a bright background and light when viewed against a dark background, which can be explained by reduced activity in the on and off pathways, respectively. If RGCs that should be firing when viewing a bright field are silent, then that silence would be interpreted by the brain as a region that is less bright (i.e. relatively dark). Similarly, if cells are silent when viewing a dark field, that silence would be interpreted as being less dark (i.e. relatively bright) (Carpenter, 1972).
The elevated contrast detection thresholds throughout the arcuate relative scotoma are reminiscent of the vision loss observed in early stages of glaucoma. Such losses could reflect a uniform loss of sensitivity by all RGCs, or a complete loss of function in only a subset of the RGCs (Osborne, Wood, Chidlow, Bae, Melena & Nash, 1999). The results of the resolution perimetry allowed us to distinguish between these two hypotheses by quantifying the sampling density within the arcuate relative scotoma. Our approach uses acuity perimetry to quantitatively assess the functional loss of retinal neurons in the diseased eye using psychophysical measurements of sampling-limited visual acuity (Thibos, 1998).
According to the sampling theory of visual resolution (Thibos, 1998), the highest spatial frequency that can be represented veridically (i.e. the Nyquist frequency) is determined by the spatial density of the coarsest mosaic of retinal neurons in the retinal pathway. Frequencies above the Nyquist limit are perceived as aliases of the stimulus. (Williams, 1985) and the onset of perceptual aliasing provides a direct estimation of neural sampling density (Coletta & Williams, 1987; Thibos et al., 1987b). For the normal healthy retina, the neural resolution limit over the range of eccentricities studied here (3°–10°) is limited by the density of the cone mosaic because they are more sparse than RGCs (Curcio & Allen, 1990; Curcio et al., 1990). However, when retinal pathology sufficiently decreases RGCs numbers such that the RGCs mosaic is coarser than the cone mosaic, parafoveal visual resolution will be a measure of RGC density rather than cone density, similar to what is measured in the far periphery (Anderson & Hess, 1990; Anderson, Mullen & Hess, 1991; Coletta & Watson, 2006; Thibos, Cheney & Walsh, 1987a; Thibos, Still & Bradley, 1996).
AOSLO imaging showed that the cone mosaic was normal indicating that the loss of visual acuity measured in our experiment cannot be explained by loss of cones. The predicted resolution acuity based on measured cone density was 22 cyc/deg at N1 and 21 cyc/deg at L1. Although this prediction was satisfied in the normal visual field above the horizontal raphe (N1), measured acuity (see Table 1) was 3-fold lower (8.6 cyc/deg) inside the arcuate defect (L1). A 3-fold reduction in sampling-limited acuity implies a 3-fold reduction in linear density (cells/mm) or a 9-fold reduction in areal density (cells/mm2). Since RGC density is normally greater than cone density at the affected eccentricities, some loss of RGCs would be needed just to make them the coarsest neural array. Starting from that reduced level, a further 9-fold loss of density occurred according to the psychophysical results. We conclude, therefore, that fewer than 11% of the RGCs in that area were functionally connected to the brain when the experiment was conducted. Thus we reject the hypothesis that the CWS produced a uniform loss of sensitivity of RGCs while preserving their spatial density.
Successful mapping of the arcuate relative scotoma required the use of a target with less than 10% contrast, indicating that contrast sensitivity remained remarkably high inside the affected region. Previous perimetric work has shown that contrast threshold for a target size of 0.3° is typically about 10% in the regions of the visual field encompassed by the scotoma (Wilson, 1970). To understand how contrast sensitivity could be almost normal despite the loss of more than 90% of RGC, consider the following quantitative argument. Sampling-limited acuity at test location L1 was 8.6 cyc/deg (Table 1), which implies a functional density of approximately 280 receptive fields per square degree. A 0.3° target covers 0.07 square degrees of area, which is large enough to cover some 20 receptive fields of functional RGCs. This suggests the remaining RGCs were healthy, sensitive, and numerous enough to provide almost normal contrast thresholds for a 0.3° target.
Atrophy of axons at the CWS site and within the arcuate region is evident in the thinning of the RNFL as seen with SDOCT (Fig. 7 and and8).8). This observation agrees with a recent report of decreased RNFL thickness in the CWS region (Alencar et al., 2007). Our data, therefore, support the hypothesis that subsequent to a local perfusion failure, axonal transport is compromised resulting in CWS formation within hours or days of the perfusion loss (Dollery et al., 1966). Interruption of axoplasmic flow has also been shown to precipitate rapid changes in the cytoskeleton of RGC axons in monkey retina (Fortune, Cull & Burgoyne, 2008; Fortune, Wang, Cull & Cioffi, 2008). These events are followed by axon atrophy proximal to the site of interruption that may take months to complete (Quigley, Hohman, Addicks & Green, 1984; Whitmore, Libby & John, 2005). Some axon atrophy is evidently permanent since we observed it 4 years after the initial vascular event (Fig. 7 & 8). An arcuate darkening of the RNFL following presumed interruption of axoplasmic flow has been reported previously in both monkeys (Quigley et al., 1984) and humans (Alencar et al., 2007). The similar observation in our patient (Fig. 6) is clearly associated with axon atrophy (Fig. 7 & 8). Interestingly, although the vascular event that precipitates the CWS may be restricted to the inner retina, our SDOCT images indicate thickening of the outer nuclear layer at the CWS site. The total retinal thickness remained relatively constant. That is, as the RNFL shrank the ONL expanded.
Experimental studies in the rat show the b-wave of the electro-retinogram drops to almost zero within one minute following ischemia (Bui, Vingrys & Kalloniatis, 2003). In humans vision loss follows within seconds after loss of perfusion in the retinal vasculature (Rubin & Walls, 1965). The visual consequences of such experimentally induced perfusion failure mimic the visual phenomena associated with pathological perfusion failures (Misra, Flanagan & Martin, 2008) and indicates that the arcuate visual anomaly reported by our patient likely reflects a rapid loss of function within the inner retina which occurred within seconds of the perfusion failure that ultimately led to the CWS and vision loss. Animal studies show that relatively short periods of perfusion loss (60–90 minutes) can result in irreversible damage of most of the RGCs (Selles-Navarro, Villegas-Perez, Salvador-Silva, Ruiz-Gomez & Vidal-Sanz, 1996; Ulrich & Reimann, 1986) indicating that within hours after the initial event our patient had suffered permanent damage. That is, by the time a CWS has developed (Dollery et al., 1966), permanent damage has probably already occurred. However, both animal studies of retinal ischemia (Hayreh & Weingeist, 1980; Hayreh, Zimmerman, Kimura & Sanon, 2004) and humans with CWSs (King-Smith et al., 1987; King-Smith, Vingrys, Benes, Grigsby & Billock, 1989) including our patient has reported recovery of visual function. The mechanism of recovery is unknown. The animal studies indicate that more prolonged ischemia increases the chance of permanent vision loss. In our patient, although there was some recovery, permanent vision loss was observed in the focal and arcuate scotomas indicating that although some axons are able to recover signal transmission, many RGCs suffer permanent damage.
The functional and structural changes that accompany a CWS support the local hypoperfusion hypothesis (McLeod, 2005). The ischemia leads to failed signal transmission as well as failed axoplasmic flow in the hypoperfused RGC fibers. The ischemia also produced a permanent loss of function and reductions in RNFL thickness, presumably due to cell death of the RGCs located at the CWS site and of the RGCs that send axons through the hypoperfused region. Some axons passing through the CWS site seemed to survive the vascular event and for some the loss of function was temporary.
This work was supported by NIH grant R01-EY05109 to LNT and NIH grants R01-EY14375 and R01-EY04395 to SAB. We thank Dr. Victor E. Malinovsky and Dr. Hua Gao of the Indiana University School of Optometry for clinical consultations and Ms. Michelle Cornett from the Indiana University Community Eye Care Center for supplying the clinical data and fundus images. We also thank Irit Zakaim and Hongxin Song for assistance with data collection. Special thanks to Professor W.R. Levick of the Australian National University for his insights into the neurophysiological implications of retinal ischemia.