ABSTRACT
A method of extracting a temporally bounded component of a composite signal has been developed which minimizes data corruption in signal processing. The composite signal is windowed in the time domain, padding signals are attached, and finally, the conditioned signal is filtered to extract the component of interest. The method has been utilized to extract the Oscillatory Potential (OP) from the Electroretinogram (ERG). ERGs can contain impulse like transients, including flash artifacts and a-b wave transition, which may not be related to the Oscillatory Potential. Such transients will stimulate a filter, yielding its natural (filter) response and thus distort the actual OP signal. To avoid this effect, time-domain windowing and signal conditioning is used to extract the OP from the ERG. The extraction and modeling approach is applied to ERGs obtained from patients with recent monocular central retinal vein occlusion (CRVO). Model parameters clearly differentiate affected from fellow eyes and show subtle differences between eyes with benign and complicated outcomes.
