Relationship of Oscillatory Potential Amplitude to A-Wave Slope Over a Range of Flash Luminances in Normal Subjects

Transcription

1 Investigative Ophthalmology & Visual Science, Vol. 32, No. 5, April 1991 Copyright Association for Research in Vision and Ophthalmology Relationship of Oscillatory Potential Amplitude to A-Wave Slope Over a Range of Flash Luminances in Normal Subjects Michael A. Sandberg,* Hokyung Lee,* G. Philip Matthews, f and Alexander R. Gaudio* The effect of flash luminance on the relationship of oscillatory potential (OP) amplitude, as a measurement of inner retinal function, to a-wave slope, as a measurement of photoreceptor function, was evaluated in full-field electroretinograms recorded from normal subjects. The ratio of OP amplitude to a-wave slope was found to be independent of flash luminance over a 3000-fold luminance range. This finding raises the possibility that a reduction in the ratio of OP amplitude to a-wave slope in eyes with media opacities may be used as a sign of inner retinal malfunction. Selected cases are presented to illustrate application of this approach. Invest Ophthalmol Vis Sci 32: ,1991 Oscillatory potentials (OPs) are wavelets that appear on the ascending limb of the b-wave in electroretinographic (ERG) responses to bright flashes and are thought to reflect feedback synaptic modulation of bipolar cell responses. 1 ' 2 In some patients with diabetic retinopathy 3 " 9 or central retinal vein occlusion, 410 these wavelets have been reported to be reduced in amplitude, without alteration of the a-wave, which is generated by photoreceptors. 11 ' 12 Since OPs are more susceptible than the a-wave to clamping of the retinal circulation, 13 their reduction in patients with a normal a-wave has been interpreted as a sign of possible inner retinal ischemia. Patients with diabetic retinopathy or central retinal vein occlusion frequently have small pupils and media opacities 14 " 17 (cataracts, vitreous or subhyaloid hemorrhages, and retinal hemorrhages of variable extent) that may diminish the amount of light reaching the photoreceptors. Vitreous opacities, for example, can attenuate light up to 1000-fold. 18 The presence of such "filters" in front of the photoreceptors makes it difficult to quantify the degree of inner retinal mal- From the *Berman-Gund Laboratory for the Study of Retinal Degenerations, Harvard Medical School, Massachusetts Eye and Ear Infirmary, Boston, Massachusetts, and the fdepartment of Ophthalmology, The New York Hospital-Cornell University Medical Center, New York, New York. Supported by National Eye Institute project grant EY00169 (MAS) and a grant from Inje University Medical College, Seoul, Korea to Harvard Medical School (HL). Submitted for publication October 27,1989; accepted November 2, Reprint requests: Dr. Michael A. Sandberg, Berman-Gund Laboratory, Massachusetts Eye and Ear Infirmary, 243 Charles Street, Boston, MA function, even when OPs are detectable with bright flashes, because attenuation of the light reduces not only these responses, but also the a-wave on which they depend. 19 " 21 We tried to determine whether (in normal subjects) a relationship could be denned between OP amplitude, as a measurement of inner retinal function, and a-wave slope, as a measurement of photoreceptor function, 2223 that would take into account or even be independent of flash luminance. A-wave amplitude was not used as a measurement of photoreceptor function, because it is determined, in part, by onset of the b-wave, 24 which is delayed with diminished retinal illumination. 20 ' 25 The OPs were quantified both in the time domain, by which amplitude has been conventionally assessed, 4 " 7 ' 26 and in the frequency domain, by which mid frequency (ie, Hz) and high-frequency (ie, Hz) bands have been separated, 27 " 29 to determine to what extent either measurement was related to the a-wave slope. Materials and Methods Basic Studies The primary subjects were six healthy volunteers (age range, years) who had normal ophthalmic examinations and <6 diopters of myopia. Informed consent was obtained from these and other subjects before examination. The pupil of one eye of each subject was dilated with 2.5% phenylephrine hydrochloride and 1% cyclopentolate hydrochloride, and the eye was dark-adapted for 45 min. Full-field ERGs were then elicited routinely in darkness with xenon flashes of white light ranging in ascending integrated luminance from to 2.34 log ft.l.-sec in 0.5-log 1508

2 No. 5 RELATION OF OP AMPLITUDE TO A-WAVE SLOPE / Sondberg er ol 1509 unit steps and presented once per minute, one flash per step. These flashes were generated by a brightflash stimulator (L.K.C., Gaithersburg, MD) positioned above a diffuser in a Ganzfeld dome and were varied in luminance with gelatin neutral-density filters (Wratten #96, Eastman Kodak Company, Rochester, NY). In pilot experiments, OPs in response to the first flash of a series of four of equal luminance, presented at 1-min intervals after 45 min of dark adaptation, were found to be,smaller than those in the subsequent three responses for luminances > 1.34 log ft.l.-sec. In other words, for this luminance range, the first flash conditioned the response to the second flash, and the second and third flashes had no additional effect on the subsequent responses. 30 Over the four responses, a-waves showed no change. 30 A conditioning effect on OPs in a succeeding response was seen even when the luminance of the first of two flashes was 0.5 log unit lower than the luminance of the second flash. Therefore, in our routine protocol of singleflashesof ascending luminance at 1-min intervals, OPs to luminances S: 1.34 log ft.l.-sec were considered to be conditioned by prior flashes. Conditioning of OPs elicited with bright flashes has been used to maximize OP amplitudes of patients with diabetic retinopathy. 6 Responses were monitored with a bipolar Burian- Allen contact lens electrode (Hansen, Iowa City, IA) coated with 1% methycellulose and placed on the cornea, topically anesthetized with 0.5% proparacaine hydrochloride. A ground electrode was placed on the ' forehead. Fixation was facilitated by a red light-emitting diode in the rear of the dome. Responses were amplified differentially at a gain of 1000 (-3 db at 2 Hz and 1000 Hz) and digitized at a sampling rate of 521 Hz over a time window of 492 msec by an analogto-digital data acquisition module (MacADIOS 411; G. W. Instruments, Cambridge, MA). Waveforms were then saved to a diskette. After testing, responses were displayed and analyzed by computer (Macintosh SE; Apple, Cupertino, CA). A-wave slopes, quantified as amplitude from initial baseline to the major cornea-negative peak divided by the time interval between a-wave onset and p^ak, were then calculated after the examiner had designated these locations with a screen cursor. Although the photoflash unit generated a variable-sized photovoltaic artifact that partially obscured the a-wave, this artifact was eliminated from each ERG response by subtracting a digitized replica of the artifact automatically scaled in amplitude to that present in the initial 2 msec of the response (Fig. IA). The OP amplitudes were quantified in the time domain between the a-wave peak (or, if absent, onset of the b-wave) and major cornea-positive b-wave peak, after excluding from analysis any responses with blink artifacts on or preceding the b-wave peak. Quantitation of OPs was done even for those responses in which OPs could not be detected by inspection, in which cases the values were considered measures of noise. First, low-frequency components were reduced with a 62 Hz high-pass digital filter. Then, the mean of the absolute value of the amplitude variation above and below baseline for the several wavelets between the a-wave and b-wave peaks was computed (Fig. 1B). The OP amplitudes were quantified in the frequency domain by next assigning zeros to points preceding the a-wave peak and following the b-wave peak and then multiplying the high-pass-filtered waveform between these peaks by a Hamming window, 31 y'(t) = y(t) ( (1-cos (2TT (t-t a )/(t b -t a )))), where t a and t b are a-wave and b-wave implicit times, respectively. The Hamming window was used to minimize voltage discontinuities preceding the peak of the a- wave and following the peak of the b-wave. A magnitude fast Fourier transform (FFT) was then done on the entire wave form, which generally revealed two distinct bands for the OPs, one peaking in the mid frequencies between Hz and the other peaking in the high frequencies between Hz (Fig. 1C). The amplitude of each of these peaks, when present, was then determined. Time- and frequency-domain OP amplitudes, derived from ERG responses to the standard luminance series, were related to the respective a-wave slopes by linear regression. A similar analysis was applied to the b-wave, generated by Miiller cells 32 and whose amplitude can also be reduced preferentially compared with the a-wave amplitude in central retinal vein occlusion. 33 " 37 B-wave amplitudes were quantified from the a-wave peak to the major cornea-positive peak and then related to a-wave slopes. Clinical Studies Since one of the analyses revealed a potentially useful measurement of inner retinal function independent of flash luminance, this measurement was then evaluated in a patient with a dense vitreous hemorrhage and presumed normal retinal function and in a patient with central retinal vein occlusion and clear media; the values were compared with values obtained from 26 normal subjects (age range, years). For these clinical studies, ERGs were elicited with our standard diagnostic protocol for patients with known or suspected retinal vascular disease, which includes dim blue (X max = 440 nm) flashes of -3.8 log ft.l.-sec to elicit rod-isolated responses, dim whiteflashesof 1.2 log ft.l.-sec to elicit mixed conerod responses, 30-Hzflashesof the same white light to

3 1510 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1991 Vol. 32 BRIGHT-FLASH ERGS WITH BIPOLAR BURIAN-ALLEN LENS ELECTRODE PHOTOVOLTAIC ARTIFACT RECORDING MINUS ARTIFACT QUANTIFICATION OF TIME-DOMAIN OSCILLATORY POTENTIAL AMPLITUDE 400 HV QUANTIFICATION OF FREQUENCY-DOMAIN OSCILLATORY POTENTIAL AMPLITUDE HIGH-FREQUENCY PEAK Fig. 1. ERGs recorded with a bipolar Burian-Allen contact lens electrode from a normal subject in response to a full-field flash of 1.85 log ft L-sec before (upper trace) and after (lower trace) subtraction of a scaled, digitized record of the photovoltaic artifact. Traces begin at flash onset. Arrow designates peak of photovoltaic artifact. Asterisk denotes blink artifact (A). Quantification of oscillatory potential amplitude in the time domain between the a-wave and b-wave peaks after 62 Hz highpass filtering for a normal subject in response to the same flash. Vertical lines represent some of the amplitudes whose absolute values with respect to baseline were summed prior to calculation of the mean (B). Quantification of mid- and high-frequency amplitudes for oscillatory potentials by the magnitude fast Fourier transform for a normal subject in response to the sameflash (C). MID-FREQUENCY PEAK / 2 ^ nv elicit cone-isolated responses, and bright white flashes of 1.85 log ft.l.-sec to elicit OPs. The responses to dim blue flashes were used to derive an estimate of OP noise. Basic Studies Results Representative ERGs show that a-waves and, to a lesser extent, OPs were detectable for a flash of log ft.l.-sec, and both appeared to increase in amplitude with increasing flash luminance (Fig. 2). At low luminances, the a-wave consisted of only a single deflection, presumably representative of rod function, whose implicit time decreased with increasing luminance. For a middle range of luminances, the a-wave consisted of a minor followed by a major deflection, thought to be indicative of superimposed cone and rod activity, respectively. 38 ' 39 For higher luminances, the a-wave became a single peak again, although both cone and rod contributions presumably remained but now had indistinguishable implicit times. The b-wave was detectable at a lower luminance than the a-wave or OPs, increased in amplitude with increasing flash luminance, and reached a maximum amplitude for the highest luminances. Linear regression of OP amplitude on a-wave slope, based on the luminance range for which a-waves were detectable, revealed significant relationships for each subject (Table 1 and Fig. 3). Mean OP amplitude, minus the y-intercept for which a-wave slope was zero, equaled 24% of a-wave slope (Fig. 3). The y-intercept (ie, 10 IAV) equaled the upper 95% limit for noise based on amplitudes calculated from those responses (n = 16) to lower luminanceflashesfor which a-waves and OPs were not detectable. This noise limit was subtracted from each OP amplitude for responses to higher luminances in which a-waves were detectable to derive the residual signal. The ratio of residual OP amplitude to a-wave slope was found to be independent of luminance for integrated luminances > -1.2 log ft.l.-sec (Fig. 4).

4 No. 5 RELATION OF OP AMPLITUDE TO A-WAVE SLOPE / Sondberg er ol 1511 INTEGRATED LUMINANCE (LOG FT.L.-SEC.) FULL-FIELD ERGS FROM A NORMAL SUBJECT uv 50 msec Fig. 2. Time domain ERGs recorded from a normal subject in response to full-field flashes of varying integrated luminance. Traces begin atflashonset. Linear regression of OP amplitude in the frequency domain on a-wave slope for each subject revealed weaker correlations than were found with time domain analysis (Table 2). On average, frequency domain OP amplitude, whether from mid- or high-frequency bands, equaled 4% of a-wave slope minus a y-intercept (Fig. 5). The higher variability seen in the frequency domain compared with the time domain was due, in part, to switching in the dominance of mid- and high-frequency peaks from one luminance step to the next. In some subjects, the "lid-frequency peak was dominant for one luminance while the highfrequency peak was dominant for the next, and vice versa. Rarely, only a single broad peak near 100 Hz was present. Based on wave forms tor which both peaks were evident, the peak high frequency was found, on average, to be the second harmonic of the peak mid frequency (Fig. 6). In contrast to the linear relationships between OP amplitude and a-wave slope, a logarithmic (le, nonlinear) relationship was found between b-wave amplitude and a-wave slope (Fig. 7). B-wave amplitude had a value of ~310 nv for an a-wave slope approaching zero and appeared to saturate for the three highest flash luminances while a-wave slope continued to increase. Clinical Studies Based on our clinical series of recordings in 26 normal subjects, we set out to define normal confidence limits for the ratio of OP amplitude to a-wave slope and for the maximum interocular percentage difference in the ratio. The ratio was found to be inversely correlated with age (r = , P =? 0.04, based on logarithmic data) as with many other absolute measures of ERG function, which prevented a practical estimation of confidence limits from this sample size. However, the interocular percentage difference in the ratio (ie, 1 - (smaller ratio -r larger ratio)) showed no correlation with age (r = 0.005; P, not significant, based on logarithmic data), and a 95% confidence limit of 32% could be derived. With respect to this normal limit, two case studies illustrate the potential application of quantifying the ratio of OP amplitude to a-wave slope to evaluate inner retinal function. The first was a 25-year-old woman with a dense vitreous hemorrhage in her left eye (OS). The patient had a 3-year history of chronic uveitis OS treated with topical and periocular corticosteroids. After a sub-tenon's injection, she had a sudden decrease in vision OS. Findings OS included hand-motions acuity, posterior synechiae limiting maximal dilation to 4 mm, and slight posterior subcapsular cataractous changes. The right eye was normal. B-scan ultrasonography revealed dispersed vitreous echoes and an attached retina. Her ERG responses to all four stimuli of our diagnostic series were subnormal and delayed OS relative to herrighteye (OD). However, the ratio of OP amplitude to a-wave slope derived from her bright-flash ERGs was actually 17% smaller OD (0.79) than OS (0.95) (Fig. 8), an insignificant interocular percentage difference compatible with a diagnosis of preserved inner retinal function OS. The second case was a 77-year-old woman with loss of vision OD secondary to central retinalvein occlu- Table 1. Time domain regression of oscillatory potential amplitude on A-wave slope* Subject A B C DE F Regression slope Y-intercept * Based on data illustrated in Figure 3. Correlation P-value

5 1512 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1991 Vol. 32 LINEAR REGRESSION OF TIME-DOMAIN OSCILLATORY POTENTIAL AMPLITUDE ON A-WAVE SLOPE IN NORMAL SUBJECTS Fig. 3. Linear regression of time domain oscillatory potential amplitude on a- wave slope for each of six normal subjects and for the mean. Values were obtained for flashes of integrated luminance between 1.66 and 2.34 log ft L-sec. Regression line for the mean is y = 0.24x , r = 0.99, P < See Table 1 for regression coefficients and statistics for each of the six subjects. A-WAVE SLOPE ((iv/msec) MEAN RATIO OF TIME-DOMAIN OSCILLATORY POTENTIAL AMPLITUDE TO A-WAVE SLOPE VS LUMINANCE IN NORMAL SUBJECTS LOG INTEGRATED LUMINANCE (FT.L-SEC.) Fig. 4. Plot of mean ratio of time domain oscillatory potential amplitude, corrected for noise (see text), to a-wave slope versus log integrated flash luminance based on six subjects. Error bars designate ± 1 SEM. Solid line is linear regression of the ratio on log luminance, y = x , r = 0.16, P = NS. The ratio for -1.7 log ft L-sec, at which the a-wave wasfirstdetectable, fell significantly below the values for higher luminances (not illustrated). sion diagnosed 1 month before ERG testing. Her medical history was notable for Graves's disease and hypertension for which she was receiving levothyroxine, enalapril, and furosemide. Visual acuity was counting fingers OD and 20/30+ OS. Slit-lamp examination revealed slight nuclear and cortical haze OU. Ophthalmoscopy showed extensive intraretinal hemorrhages with dilated tortuous veins, macular hemorrhages, and edema OD and slight arterioiar narrowing OS. Fluorescein angiography was interpreted as inconclusive as to the extent of ischemia. Her ERG responses were reduced and delayed to all four stimuli OD relative to OS (Fig. 9). The ratio of OP amplitude to a- wave slope derived from her responses to bright flashes was 0.21 OD and 0.40 OS. Her 47% reduction OD relative to OS for the ratio fell outside our normal 95% confidence limit of 32%. Therefore, this patient's recordings are compatible with a diagnosis of inner retinal malfunction.

6 No. 5 RELATION OF OP AMPLITUDE TO A-WAVE SLOPE / Sondberg er ol 1513 Table 2. Frequency domain regression of oscillatory potential amplitude for mid- (top) and highfrequency (bottom) peaks on A-wave slope PEAK HIGH-FREQUENCY DIVIDED BY.PEAK MID-FREQUENCY FOR OSCILLATORY POTENTIALS IN NORMAL SUBJECTS Subject Regression slope Y-intercept Correlation P-value A B C D E F ' > n.s n.s RATIO OF PEAK HIGH-FREQUENCY + PEAK MID-FREQUENCY Fig. 6. Histogram of peak high-frequency H- peak mid-frequency for waveforms showing both mid- and high-frequency bands based on six subjects (n = 38). The mean ratio was 2.04 ± Discussion Our study shows that, in normal subjects, OP amplitude was proportional to a-wave slope for responses toflashesof varying luminance. On average, the ratio of OP amplitude quantified in the time domain to a-wave slope was independent of flash luminance from the highest luminance down to a 3.5 log-unit or ~ 3000-fold attenuation of luminance. Since OP amplitude is a measurement of inner retinal function, 12 LINEAR REGRESSION OF FREQUENCY-DOMAIN OSCILLATORY POTENTIAL AMPLITUDES ON A-WAVE SLOPE IN NORMAL SUBJECTS and the a-wave slope is a measurement of photoreceptor function, 11>12 ' 22 ' 23 the constancy of this ratio over a wide range of flash luminances in normal subjects with clear media raises the possibility that an abnormally low ratio in patients with media opacities might be a reason to suspect that they have inner retinal malfunction. Examination of two patients with unilateral disease using our diagnostic ERG protocol consisting of dim flashes to isolate cone and rod components and bright flashes to elicit OPs supported this idea. One patient with a dense vitreous hemorrhage had comparable ratios in her two eyes despite having subnormal and delayed responses to dim flashes and subnormal OPs to brightflashesin her affected eye. Thisfindingcould be explained by an attenuation of light incident on normal outer and inner retina. Conversely, a patient with central retinal vein occlusion and relatively clear media had a significantly smaller ratio, abnormal responses to dim flashes, and subnormal OPs in her B-WAVE AMPLITUDE VS A-WAVE SLOPE IN NORMAL SUBJECTS MEAN * # A-WAVE SLOPE (nv/msec) Fig. 5. Linear regression of oscillatory potential amplitudes for mid- and high-frequency peaks on a-wave slope for mean of six subjects. Values were obtained forflashesof integrated luminance between and 2.34 log ft L-sec. Regression lines are y = x , r = 0.94, P < for the mid-frequency peak and y = 0.044x , r = 0.98, P < for the high-frequency peak. See Table 2 for regression coefficients and statistics for each of the six subjects * A-WAVE SLOPE (uv/msec) Fig. 7. Plot of mean of b-wave amplitude versus a-wave slope based on six subjects. Values were obtained forflashesof integrated luminance between and 2.34 log ft L-sec. The data could be fitted by y = log x , r = 0.99, P < (not illustrated).

7 1514 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / April 1991 Vol. 32 CASE STUDY: VITREOUS HEMORRHAGE OS 00 BLUE WHITE BRIGHT-FLASH WHITE Fig. 8. ERG responses from a patient with a vitreous hemorrhage OS. Responses were elicited after 45 min of dark adaptation and two or three consecutive traces are superimposed. Oscillatory potential noise was derived from the average response to blue light. For the bright-flash recordings, asterisks designate the point on each b-wave up to which oscillatory potentials were quantified in order to exclude blink artifacts. Quantifications of oscillatory potential amplitude and a-wave slope were also based on the average waveform. For further details of methods, see text. CASE STUDY: CENTRAL RETINAL VEIN OCCLUSION OD 00 OS BLUE Fig. 9. ERG responses for a patient with central retinal vein occlusion OD and relatively clear media. For details of recording and analysis, see Figure 8 legend and text. 30 HZ WHITE BRIGHT-FLASH WHITE

8 No. 5 RELATION OF OP AMPLITUDE TO A-WAVE SLOPE / Sondberg er ol 1515 affected eye. This finding was consistent with inner retinal malfunction and, possibly, ischemia. The OP amplitudes quantified by Fourier analysis were also linearly proportional to a-wave slope, but the correlation coefficients for mid- and high-frequency peaks were smaller than those obtained from time-domain analysis. This resulted from the fact that, for the mid-frequency or high-frequency peak, FFT amplitude grew less systematically than time-domain amplitude with increasing luminance, probably due to shifting dominance between the two frequency peaks. Since the high-frequency peak averaged the second harmonic of the mid-frequency peak, both could reflect the output of the same retinal oscillators with the capacity for intermittent, variable full-wave rectification. 40 In other words, the two frequency bands may arise from interdependent sources. This is supported by the finding that the ratios of both mid- and high-frequency peak amplitudes to a-wave slope were the same. Alternatively, since both cone and rod components were present in the a-wave, it is possible that shifting dominance may reflect a luminance-sensitive interaction between these two contributions to the OPs. In contrast to OP amplitude, b-wave amplitude was not linearly correlated with a-wave slope. For high flash luminances, b-wave amplitude became stable and relatively independent of a-wave slope. This nonlinear relationship indicates that b-wave amplitude relative to a-wave slope cannot serve as a useful index of inner retinal malfunction in patients with dense media opacities. For example, patients could have inner retinal ischemia, which may preferentially reduce b-wave amplitude, and a vitreous hemorrhage, which will preferentially reduce a-wave slope, resulting in a normal value for the ratio of b-wave amplitude to a- wave slope. Moreover, the large y-intercept (ie, 310 jiv) represents a second parameter that would have to be estimated for each patient. On the other hand, with the brightest flash used in this study, b-wave amplitude would be insensitive to opacities that attenuate light < tenfold and, in such cases, could be used by itself to assess inner retinal malfunction. Digital subtraction of the photovoltaic artifact from each response before quantifying a-wave slope permitted use of a conventional bipolar Burian-Allen contact lens electrode with its unshielded metallic ring to record bright flash responses. This approach was adopted as a convenient alternative to developing a custom monopolar nonmetallic or light-shielded metallic electrode to minimize or prevent a photovoltaic artifact in bright flash recordings. 41 " 43 Analysis of OPs only between the a-wave and b-wave peaks eliminated the problem of additional, erroneous amplitude due to blink artifacts subsequent to the b-wave peak, which might lie in the same frequency bands as OPs. Quantitation of OP amplitude as the mean absolute value of the area between each wavelet and baseline, unlike the "caliper-square" method of summing peak amplitudes, 4 did not depend on the number of peaks thought to be present in the response. Since decreased retinal illumination appears not to alter the ratio of OP amplitude to a-wave slope based on our findings in normal subjects and a patient with a vitreous hemorrhage, it would be interesting to study this ratio in patients with known inner retinal ischemia with or without intraretinal hemorrhages. Initially this work should be done in patients with presumed unilateral disease, since a normal confidence limit for the ratio has been defined up to now only with respect to an interocular comparison. The ratio should be abnormally low in the affected eyes but comparable in the two groups of patients, assuming that the intraretinal blood has not, itself, been toxic to 44 or indirectly affected 45 the outer or inner retina. If confirmed, the ratio of OP amplitude to a- wave slope may then be used in patients with presumed unilateral central retinal vein occlusion to evaluate possible inner retinal ischemia even in the presence of media opacities. If and when age-corrected limits have been set for normal eyes considered in isolation, then this approach also could be applied to both eyes of patients with diabetes mellitus. Key words: a-wave, electroretinography, oscillatory potentials, retinal vascular disease, vitreous hemorrhage References 1. Ogden TE: The oscillatory waves of the primate electroretinogram. Vision Res 13:1059, Heynen H, Wachtmeister L, and van Norren D: Origin of the oscillatory potentials in the primate retina. Vision Res 25:1365, Yonemura D, Aoki T, and Tsuzuki K: Electroretinogram in diabetic retinopathy. Arch Ophthalmol 68:19, Algvere P: Clinical studies on the oscillatory potentials of the human electroretinogram with special reference to the scotopic b-wave. Acta Ophthalmol (Copenh) 46:993, Simonsen SE: The value of the oscillatory potential in selecting juvenile diabetics at risk of developing proliferative retinopathy. Metab Pediatr Syst Ophthalmol 5:55, Bresnick GH, Korth K, Groo A, and Palta M: Electroretinographic oscillatory potentials predict progression of diabetic retinopathy: Preliminary report. Arch Ophthalmol 102:1307, Kawasaki K, Yonemura K, Yokogawa Y, Saito N, and Kawakita S: Correlation between ERG oscillatory potential and psychophysical contrast sensitivity in diabetes. Doc Ophthalmol 64:209, Bresnick GH and Palta M: Predicting progression to severe proliferative diabetic retinopathy. Arch Ophthalmol 105:810, 1987a. 9. Bresnick GH and Palta M: Oscillatory potential amplitudes: Relation to severity of diabetic retinopathy. Arch Ophthalmol 105:929, 1987b.

9 1516 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1991 Vol Gaudio AR, Lee H, Kini M, Sandberg MA, and Berson EL: Oscillatory potentials in central retinal vein occlusion. ARVO Abstracts. Invest Ophthalmol Vis Sci 30(Suppl):477, Brown KT and Wiesel TN: Localization of origins of electroretinogram components by intraretinal recording in the intact cat eye. J Physiol (Lond) 158:257, Brown KT and Watanabe K: Isolation and identification of a receptor potential from the pure cone fovea of the monkey retina. Nature 193:958, Brown KT: The electroretinogram: Its components and their origin. Vision Res 8:633, Hayreh SS, Rojas P, Podhajsky P, Montague P, and Woolson RF: Ocular neovascularization with retinal vascular occlusion: III. Incidence of ocular neovascularization with retinal vein occlusion. Ophthalmology 90:488, Benson WE, Tasman W, and Duage TD: Diabetic retinopathy. In Clinical Ophthalmology, Vol 3, Duane TD and Jaeger EA, editors. Philadelphia, JB Lippincott, 1988, pp Yanoff M and Fine BS: Ocular Pathology. Philadelphia, Harper & Row, 1989, pp Sanborn GE, Magargal LE, and Jaeger EA: Venous occlusive disease of the retina. In Clinical Ophthalmology, Vol 3, Duane TD and Jaeger EA, editors. Philadelphia, JB Lippincott, 1988, pp Knighton RW and Blankenship GW: Electrophysiologic evaluation of eyes with opaque media. In Electrophysiology and Psychophysics: Their Use in Ophthalmic Diagnosis, Sokol S, editor. Boston, Little, Brown, Intern Ophthalmol Clinics 20:1, 1980, pp Cone RA: The early receptor potential of the vertebrate eye. Cold Spring Harb Symp Quant Biol 30:483, Fuller DG, Knighton RW, and Machemer R: Bright-flash electroretinography for the evaluation of eyes with opaque vitreous. Am J Ophthalmol 80:214, Perlman I: Relationship between the amplitudes of the b wave and the a wave as a useful index for evaluating the electroretinogram. Br J Ophthalmol 67:443, Fulton AB and Rushton WAH: The human rod ERG: Correlation with psychophysical responses in light and dark adaptation. Vision Res 18:793, Van Norren D and Valeton JM: The human rod ERG: The dark-adapted a-wave response function. Vision Res 19:1433, Siegel IM and Carr RE: Electrodiagnostic and psychophysical testing in retinal disease. In Electrophysiology and Psychophysics: Their Use in Ophthalmic Diagnosis, Sokol S, editor. Boston, Little, Brown, Intern Ophthalmol Clinics 20:21, 1980, pp Sandberg MA, Sullivan PL, and Berson EL: Temporal aspects of the dark-adapted cone a-wave in retinitis pigmentosa. Invest Ophthalmol Vis Sci 21:765, Yonemura D and Hasui I: A statistical study of the oscillatory potential of the normal human electroretinogram. Folia Ophthalmologica Japonica 22:260, Algvere P and Westbeck S: Human ERG in response to double flashes of light during the course of dark adaptation: A Fourier analysis of the oscillatory potentials. Vision Res 12:195, Gur M and Zeevi Y: Frequency-domain analysis of the human electroretinogram. J Opt Soc Am [A] 70:53, Van der Torren K, Groenweg G, and Van Lith G: Measuring oscillatory potentials: Fourier analysis. Doc Ophthalmol 69:153, Wachmeister L: On the oscillatory potentials of the human electroretinogram in light and dark adaptation: IV. Effect of adaptation to shortflashesof light: Time interval and intensity of conditioning flashes: A Fourier analysis. Acta Ophthalmol (Copenh) 50:250, Ramirez RW: The FFT: Fundamentals and Concepts. Englewood, NJ, Prentice Hall, 1985, p Miller RF and Dowling JE: Intracellular responses of the Miiller (glial) cells of mudpuppy retina: Their relation to b- wave of the electroretinogram. J Neurophysiol 33:323, Henkes HE: Electroretinography in circulatory disturbances of the retina: I. Electroretinogram in cases of occlusion of central retinal vein or of one of its branches. Arch Ophthalmol 49:190, Karpe G and Uchermann A: The clinical electroretinogram: VII. The electroretinogram in circulatory disturbances of the retina. Acta Ophthalmol (Copenh) 33:493, Sabates R, Hirose T, and McMeel JW: Electroretinography in the prognosis and classification of central retina vein occlusion. Arch Ophthalmol 101:232, Kaye SB and Harding SP: Early electroretinography in unilateral central retinal vein occlusion as a predictor of rubeosis iridis. Arch Ophthalmol 106:353, Breton ME, Quinn GE, Keene SS, Dahman JC, and Brucker AJ: Electroretinogram parameters at presentation as predictors of rubeosis in central retinal vein occlusion patients. Ophthalmology 96:1343, Armington JC, Johnson EP, and Riggs LA: The scotopic a- wave in the electrical response of the human retina. J Physiol (Lond) 118:289, Auerbach E and Burian HM: Studies on the photopic-scotopic relationships in the human electroretinogram. Am J Ophthalmol 40:42, Cornsweet TN: The Design of Electric Circuits in the Behavioral Sciences. New York, John Wiley and Sons, 1963, p Berson EL and Goldstein EB: The early receptor potential in dominantly inherited retinitis pigmentosa. Arch Ophthalmol 83:412, Barber C, Cotterill DJ, and Larke JR: A new contact lens electrode. Documenta Ophthalmologica Proceedings Series 13:385, Sieving PA, Fishman GA, and Maggiano JM: Corneal wick electrode for recording bright flash electroretinograms and early receptor potentials. Arch Ophthalmol 96:899, Mandelbaum S, Cleary PE, Ryan SJ, and Ogden TE: Brightflash electroretinography and vitreous hemorrhage: An experimental study in primates. Arch Ophthalmol 98:1823, Ehrenberg M, Thresher RJ, and Machemer R: Vitreous hemorrhage nontoxic to retina as a stimulator of glial and fibrous proliferation. Am J Ophthalmol 97:611, 1984.