Evoked potential analysis: on-line signal optimization using a mini-computer

An automated computer-based system is described for the analysis of evoked potentials. All procedures are carried out in real-time A small computer performs the following functions; timing of random stimulus presentation, rejection of artifact contaminated responses, collection of digital data, computation of averaged evoked responses, computation of the Wiener filter, storage of the filtered and unfiltered averages and display of the resultant averages. The Wiener filter as described by Walter (1969) and Doyle (1975) is used to improve the estimate of the evoked potential by discriminating against frequencies likely to be contaminated with noise. The defining equation for the Wiener filter states that information at any frequency is to be weighted by the ratio of the power known to be in the signal (response) at that frequency over the corresponding power known to be in both the signal (response) and the noise (background EEG) at the same frequency. The technique requires the computation of the Fourier transform for each response in order to produce the power spectra necessary for the Wiener filter. Earlier reports dealing with this technique have usee large computers to analyze the evoked potential data off-line. The system described here allows for greater routine utilization of this powerful technique and the concomitant automated rejection of artifact contaminated responses. Highly improved estimates of the evoked potential are resultant using a minimal number of stimuli.