Automatic enhancement of single evoked potentials

Abstract

Using EEG data acquired during a simple (sRT) or choice (cRT) reaction time task we tested a signal enhancement technique. Our findings indicate that the signal enhancement technique under study substantially reduced background activity while retaining physiologically significant information. For each RT trial, a pre-stimulus epoch was modeled using a canonical variate analysis (CVA) system identification technique to produce a filter. This filter was then applied to the post-stimulus epoch to remove ongoing background activity. Signal/noise enhancement was apparent from comparison of the ensemble averages of seven raw responses and the 7 corresponding CVA-enhanced responses. The extracted single evoked potentials (sEPs) were searched by machine algorithm for an expected negative peak with approximately 200 msec latency, the latency of each identified peak estimated, and correlations between RT and peak latency computed. For the choice task significant correlations were found at the right central placement (C4) for left-handed responses and at the left parietal placement (P3) for right-handed responses. For the simple task with right-handed responses, the most significant results were found at left placements (C3 and T5) with additional significant findings bilaterally (T3, T4, P4 and O1).