Detection of stimulus-related (evoked response) activity in the electroencephalogram (EEG).

Abstract

In clinical neurophysiology the visual evaluation of averaged evoked waveforms may be rather difficult, allowing no definite answer to the question of whether stimulus-related activity is present or not. This problem calls for appropriate statistical procedures testing for the presence of significant response activity. In the present work the commonly used averaging technique was supplemented by phase value measurements of Fourier components in the poststimulus sample functions. If the latter contain no evoked activity, the phase values may be assumed to be uniformly distributed in the interval 0°-360°(the null hypothesis assumption). Contrary to this, an aggregation of phase values and, hence, a nonuniform distribution will result if evoked activity is present. The statistical problem may thus be reduced to testing a null-hypothesis assumption of uniformity. In the present investigation a statistical procedure of the Kolmogorov-Smirnov type, Kuiper's VN test, appeared to be a valuable aid in the detection of stimulusrelated EEG activity.