Analysis of the time series of the EEG frequency spectra and of EEG spectral power densities.

Some examples of the use of the principal component model for the economic description of the structure of the multiple time series and for the data reduction in the quantitative EEG studies are presented. The broad-band EEG frequency spectra were measured with the use of an electronic system designed by J. Dvorák. The EEG spectral power densities were computed via the discrete Fourier Transform (namely FFT) algorithm. The estimated two or three first principal components account for the major part of the total variance of individual EEG variables: The results hold for the used elementary epoch of measurement, i.e. 5 sec. - With the use of the algorithms and FORTRAN IV programs developed by J. Andĕl, T. Cipra and L. Tomásek a data reduction by a factor of 1:2000 can be achieved without any substantial loss of biological information. - The described methods help to obtain a better insight into the structure of the data and represent a powerful tool for data reduction at least in a certain class of experimental EEG studies (experimental toxicology, pharmacology, experimental neurology).