Postprocessing of spectral output of EEG signals for display of long records

In this paper, a linear predictive (LP) analysis of several segments of EEG data is performed and the spectral information in condensed form is displayed as position of three major peaks of smoothed LP spectrum versus time. A zero-crossing analy sis of the EEG record is also performed frame by frame so as to get a rough estimate of major frequency component in each segment. A comparison of curves of major frequency component versus time obtained by the above two methods shows a close sim ilarity between the two displays. The displays provide EEG data in a convenient condensed form for a clinician and also helps in the classification of EEC. by computer pattern recognition techniques.