A quantitative method for classification of EEG in the fetal baboon

M. M. Myers, R. I. Stark, W. P. Fifer, P. G. Grieve, J. Haiken, K. Leung and K. F. Schulze Department of Psychiatry, Columbia College of Physicians and Surgeons, New York, New York. Electroencephalographic (EEG) activity is used as a primary indicator of sleep states in adults and infants of many species and in the ovine fetus. We recently reported that the baboon fetus exhibits visually discernable patterns of EEG activity. One pattern of activity, characterized by the intermittent presence of repetitive bursts of high-voltage EEG, is indistinguishable from trace alternant (TA). TA is a distinctive pattern of EEG activity found only during early stages of development in primates. TA is the predominant pattern of EEG activity during quiet sleep in human infants < 2 mo of age. The focus of this study was to derive quantitative parameters that would discriminate TA from other activity and then to develop a method for automated categorization of EEG patterns. Results demonstrate that several parameters derived from frequency-domain analyses are related to visually coded EEG states. Among these parameters, high-frequency power (12-24 Hz) and spectral-edge frequency are good discriminators of EEG patterns. This paper describes a new parameter, EEG ratio, computed as spectral power in the rectified EEG within a band that corresponds to the frequency of bursts of activity during TA (0.03-0.20 Hz) divided by power in the 12- to 24-Hz band. This new composite parameter of EEG activity provides a markedly better correlate of visually coded EEG than any of the individual parameters tested. Using cluster analysis, we devised a method for objective minute-by-minute dichotomization of EEG ratio. The method produces results that agree with visual coding of EEG activity 87.1% of the time.