Autoregressive spectral array for graphical display of EEG data

Abstract

A graphical display of the frequency content of background electroencephalogram (EEG) activity is obtained by calculating the spectral estimates using autocorrelation autoregressive method and the classical Fourier transform method. Display of spectral content of consecutive data segments is made using hidden-line suppression technique so as to get a spectral array. The autoregressive spectral array (ASA) is found to be sensitive to baseline drift. Following baseline correction the autoregressive technique is found to be superior to the Fourier method of compressed spectral array (CSA) in detecting the transitions in the frequencies of the signal. The smoothed ASA gives a better picture of transitions and changes in the background activity. The ASA can be made to adapt to specific changes of dominant frequencies while eliminating unnecessary peaks in the spectrum. The utility of the ASA for background EEG analysis is discussed.