Artifact detection and removal during auditory evoked potential monitoring

Abstract

Various artifacts can distort or obscure evoked potential waveforms. The algorithms presented in this paper scan the output electroencephalographic signal for artifacts during evoked potential recordings. If possible, the artifact is removed; if not possible, that sweep is excluded from the averaging process required to raise the evoked response above the background electroencephalographic activity. An artifact is detected if 1 or more amplitude or frequency parameters exceed a threshold. These thresholds have been determined after constructing histograms of the parameters concerned using a number of control evoked potential recordings containing no visually recognizable artifacts. The distributions of the parameters shown by these histograms give information about their normal range. The method improves the quality of the waveform in many cases, but its effectiveness strongly depends on the characteristics of the artifacts concerned.