BICOHERENCE OF INTRACRANIAL EEG IN SLEEP, WAKEFULNESS AND SEIZURES

The hypothesis that the intracranial EEG has local structure and short -term non-stationarity is tested with a little-studied measure of nonl inear phase coupling, the bicoherence in human subdural and deep tempo ral lobe probe data from 11 subjects during sleeping, waking and seizu re states. This measure of cooperativity estimates the proportion of e nergy in every possible pair of frequency components, F-1, F-2 (from 1 to 50 Hz in this study), that satisfies the definition of quadratic p hase coupling (phase of component at F-3, Which is F-1 + F-2, equals p hase of F-1 + phase of F-2). Derived from the bispectrum, which segreg ates the non-Gaussian energy, auto-bicoherence uses the frequency comp onents in one channel; cross-bicoherence uses one channel for F-1 and F-2 and another for F-3. These higher order spectra are used in physic al systems for detection of episodes of non-linearity and transients, for pattern recognition and robust classification, relatively immune t o Gaussian components and low signal to noise ratios. Bicoherence is f ound not to be a fixed character of the EEG but quite local and unstab le, in agreement with the hypothesis. Bicoherence can be quite differe nt in adjacent segments as brief as 1.6 s as well as adjacent intracra nial electrodes as close as 6.5 mm, even when the EEG looks similar. I t can rise or fall steeply within millimeters. It is virtually absent in many analysis epochs of 17 s duration. Other epochs show significan t bicoherence with diverse form and distribution over the bifrequency plane, Isolated peaks, periodic peaks or rounded mountain ranges are e ither widely scattered or confined to one or a few parts of the plane. Bicoherence is generally an invisible feature: one cannot usually rec ognize the responsible form of non-linearity or any obvious correlate in the raw EEG. During stage II/III sleep overall mean bicoherence is generally higher than in the waking state. During seizures the diverse EEG patterns average a significant elevation in bicoherence but have a wide variance. Maximum bispectrum, maximum power spectrum, maximum a nd mean bicoherence, skewness and asymmetry all vary independently of each other, Cross-bicoherence is often intermediate between the two au to-bicoherence spectra but commonly resembles one of the two. Of the k nown factors that contribute to bicoherence, transient as distinct fro m ongoing wave forms can be more important in our data sets. This meas ure of non-linear higher moments is very sensitive to weak quadratic p hase coupling; this can come from several kinds of waveforms. New meth ods are needed to evaluate their respective contributions. Utility of this descriptor cannot be claimed before more carefully defined and re peatable brain states are studied. (C) 1997 Elsevier Science Ireland L td.