Peak component latency-corrected average method for evoked potential waveform estimation.

Abstract

A peak component latency-corrected average (PC-LCA) method consisting of three steps is presented for estimating the evoked potentials (EP) in the human brain. The present PC-LCA is essentially an alternative version of the original LCA method developed by McGillem et al. It first uses a time-varying adaptive filter as a preprocessor to suppress the ongoing EEG and to enhance the EP. Then, the statistics of mean latencies and amplitudes of the EP peak components are estimated by a peak detection and alignment procedure. The final waveform estimate is obtained by superimposing the estimated mean peak components on background information and subsequently fitting a truncated Fourier series using the fast Fourier transform (FFT). Consistent estimates of the EP peak component information are established based on asymptotic analysis of the PC-LCA. Finally, real-time implementation of the PC-LCA is considered, and the superior performance of the PC-LCA method is confirmed by detailed numerical results.