Detailed study of temporal variations of EEG by improved MEM spectral analysis with high resolution power

Abstract

The authors analyzed the electroencephalogram (EEG) time series data of a subject under general anesthesia using a newly devised method proposed by the authors. The analysis revealed an exponential characteristic in power spectral density (PSD) of the EEG, known to be peculiar to nonlinear phenomena. In addition, powers of such frequency components in EEG as alpha, beta, theta and delta were calculated quantitatively by integrating the PSD obtained from the segments of the EEG without any filtering procedures, and their temporal variations were investigated in detail. They were remarkably influenced by the stage of anesthesia. Further, the magnitude of the slope of exponentially decaying PSD increased as anesthesia became effective. The authors therefore proposed to take the magnitude as a new measure of estimating brain activities, and its usefulness was confirmed by the fact that the magnitudes for EEGs of children decrease as ages of subject increase.