Oscillatory Potential-based Characterization of the Human Light-adapted Electroretinogram Using Discrete Wavelet Transform

Abstract

Purpose: Our aim was to apply multiple discrete wavelet transformation (DWT) types to healthy light-adapted (cone) electroretinogram (ERG) signals in order to optimize DWT analysy in ERG. Oscillatory potentials (OP) were individually extracted from the signals and used to calculate an indicator for ERG analysis.
Methods: Light-adapted (LA) 3.0 cd.s/m2 ISCEV standard ERGs were recorded from both eyes of 15 healthy volunteers (mean age: 36.9 ± 13.0 years old; 13 females). LA ERG signal components, such as b-wave and OPs, were analyzed using the discrete wavelet transformation (DWT). An index (%OPi) was proposed to estimate the individual oscillatory potentials (OP1-OP5) by calculating the coefficient ratio of the OP to b-wave. Multiple mother wavelet functions (i.e., Daubechies, Symlet, and Coiflet) with five orders were applied and compared statistically using Wilcoxon tests and paired t-test comparisons with Bonferroni posthoc analyses (p < 0.005).
Results: OP4 shows the most energy at both low and high-frequency bands (80Hz and 160Hz), while OP2 has lower energy at the low-frequency band (80Hz) and higher energy at the high-frequency band (160Hz). The %OP2 is the largest among the five individual OPs. %OPs obtained with different wavelet functions differ from each other. Db2 and sym2 seem to be the optimal wavelets for analyzing light-adapted ERG components.
Conclusion: Individual OPs of the light-adapted ERG obtained with the DWT analysis may characterize different levels of retinal dysfunction. The %OPi may serve as an indicator in ERG analysis.