Discrete Wavelet Transform (DWT) Of The ERG More Accurately Predicts The End Stage Of Retinal Degenerative Disorders

Abstract

Purpose: : DWT (time-frequency domain) and, less accurately, amplitude [time-amplitude domain: TAD)] analysis of normal ERGs reveal that the progressive decay in b-wave amplitude obtained with the Photopic Hill (PH) is achieved in two asymptotical steps of 14.44±2.18µV/0.2 log-unit of flash attenuation (step1) and -1.38±0.48µV (step 2). The asymptotes intersection is normally observed in ERGs evoked to a -1.21 log cd.s.m² stimulus. We examined if knowledge of the above could help in the design of follow-up strategies for patients affected with severe outer retinal degeneration such as retinitis pigmentosa (RP).

Methods: : DWT and TAD descriptors obtained from follow-up (4 to 10 visits; monitoring period: 5 to 25 years) photopic ERGs (flash intensity: 0.64 log cd.sec.m^-2; background: 30 cd.m^-2) of patients (n=60) were plotted against the first up to the second to last follow-up ERG. In order to predict the last (known) measure (i.e. 4^th or 10^th values), longitudinal regressions were used to build pathological degeneration models, where ERGs were normalized according to the normal ERG amplitude evoked to the same flash luminance, yielding a percent-of-normal (PON) value. Prediction errors (% of error) were computed for each predicted ERG values.

Results: : DWT prediction errors [mean error: 5.94±7.12%; max: 55.84%; min: 0.29%] were significantly smaller (p<0.05) than TAD errors [mean error 23.35±31.28%; max: 88.12%; min: 5.21%]; an advantage that was found to be critical when analyzing highly corrupted signals (signal-to-noise ratio<1) such as those that often characterize the end-stage of severe degenerative retinopathies. Two distinct asymptotical rates of degeneration were observed, namely: -6.22±2.11% / year (rate1: 42% of patients) and -1.31±0.51% / year (rate2: 57% of patients). The two asymptotes intersected at a PON value of approximately 40%.

Conclusions: : Once again our results confirm the superiority of time-frequency analysis of the ERG. Similar to what was previously reported for the normal photopic ERG, with advancing retinopathy, the pathological photopic ERG also degraded following two distinct asymptotes, the difference between the two asymptotes being slightly larger in normals. Of interest outer-inner retina correlations (a-b waves correlations) also suggest that both models (Photopic Hill and progressive retinal degeneration) are nearly identical [PH: b-wave = 3.5(a-wave)+0.22µV, R²=0.94; Patho: b-wave = 2.9(a-wave)+0.22µV, R²=0.84; p>0.05]. Our finding of a biphasic degenerative process that is ERG determined should permit a more accurate staging and prognosis for these patients. Funded by FFB (USA) and Réseau-Vision.

Keywords: electroretinography: clinical • electroretinography: non-clinical • retinitis