Empirical mode decomposition denoising of the electroretinogram to enhance measurement of the photopic negative response

Abstract

Background and objectiveThe photopic negative response (PhNR) of the electroretinogram (ERG) can be useful in the diagnosis and management of glaucoma. Its utility is limited by its large degree of measurement variability, which can be caused by low frequency oscillations and baseline wander. In this work we show a new denoising strategy using the empirical mode decomposition (EMD) which improves repeatability. MethodsERGs were performed on both eyes of 56 patients using red-on-blue stimulus after 2 min of preadaptation, capturing 10 sweeps per recording. Two recordings were obtained to provide intrasession test–retest estimates. Denoising was performed by removing the residue after the decomposition with the EMD, the Ensemble EMD (EEMD) and the Complete ensemble EMD with adaptive noise (CEEMDAN). The PhNR was measured as the amplitude, relative to the prestimulus baseline, of the first negative going trough after the b wave. The coefficient of repeatability (CoR) was calculated to represent 95% of the test–retest differences for the PhNR between sessions. ResultsDenoising with all of the varieties of the EMD improved the test–retest repeatability of the PhNR. The best results were achieved with the CEEMDAN algorithm, which for the PhNR amplitude improved the coefficient of repeatability relative to the effective dynamic range from 20.2% to 8.3%. ConclusionsThe EMD can be used to remove low frequency baseline drift and oscillation from the ERG and improve the test–retest repeatability of PhNR measurements. This technique could be deployed for a range of visual and other electrophysiological tests.