Electroencephalogram Processing Method for More Accurate Determination of Cognitive Process Indicators

To study neurophysiological mechanisms of functioning of the nervous system and psyche, modern computer technologies and methods of mathematical data processing are used. One of the non-invasive methods of determining the functional state of the brain is electroencephalography. The use of the spectral singular analysis method, together with the traditional approach using Fourier analysis, is proposed to determine the indicators of cognitive processes. Computation of the singular spectrum is a rather computationally expensive process, but modern technologies allow to perform capacious computations with acceptable speed and accuracy, which allows to use new methods of data analysis. The proposed approach allows us to decompose the original signal into elementary components, which are most often described by quasi-harmonic functions. It is further proposed to perform nonparametric robust estimation of frequencies, based on which some elementary components can be grouped on the basis of close spectral characteristics. The results of calculations on the experimental data are presented, and the change of amplitude and frequency of alpha rhythm for the case of open and closed eyes of the subject is shown. As a result, the proposed method makes it possible to clarify the main characteristics of electroencephalogram signals.