Period-peak analysis of the EEG with microprocessor applications.

This paper describes the development and evaluation of a period-peak algorithm for background analysis of the clinical electroencephalogram (EEG). The procedure is a time-domain method which is harmonious with manual interpretation of the EEG tracing. Conceptually the algorithm functions in 2 modes. Major counts are detected by successive baseline crossings in the period analysis mode. Presence of superimposed activity between major-counts induces a transition to the peak-detection mode. In this manner, period-peak analysis is capable of detecting the simultaneity of slow base-waves and relatively fast superimposed activity in the EEG. Preliminary studies have been conducted in which the analysis results of this procedure were compared to those of other EEG algorithms. In general, the period-peak algorithm offered less bias towards either end of the EEG spectrum. Subsequent to testing of a FORTRAN version, the period-peak algorithm has been implemented in assembly language on a dedicated microprocessor system for on-line analysis of EEG data.