Processing visual evoked potentials based on matched filtering of single trial responses

Physiological signals recorded after presentation of a stimulus can be considered as the sum of an evoked sensory response and uncorrelated noise corresponding to the spontaneous neural background activity. In case of low signal-to-noise ratio, the sensory signal is estimated by averaging a large number of single trial records sampled under the same experimental conditions. The averaging technique, however, is inappropriate for displaying trial-to-trial variations that may occur during a recording session. A method based on matched filtering was developed for processing single visual evoked potentials (VEPs). To illustrate the performance of this method, a set of VEPs for grating patches of 8 different contrasts (including 0) was evaluated. The VEPs were subjected to signal detection analysis by computing for each contrast similarity indices between the single trial records and the averaged response used as a template. The signals were analysed in terms of probability density distributions, expressing the goodness of fit between template and each single waveform.