Ya.A. Turovskyi - Ph.D. (Мed.), Associate Professor, Head of Laboratory, Voronezh State University

The given paper presents a method to reveal the peculiarities of the EEG signal in the time-frequency domain. The method is based on the allocation of local maxima matrix W2(a,b) chains, reflecting the presence of expressed frequency components in a signal, which change their parameters (frequency and energy) in time. Frequency-limited with local minima chains, a chain of local maxima forms the feature of the signal, which may be of interest from the point of view of the interpretation of the clinical and physiological phenomenon. There are proposed two ways of limiting the peculiarities of the signal's frequency: reset all values of the matrix W2(a,b) outside the local minima, who are limiting features of the signal, and the approximation of fronts of the scalograms (local wavelet spectra) on segments from the local maximum to local minimum. Localization of the signal feature in time also has two ways of implementation: the localization of the event, to which there can be attributed both stimulation of the user, and a number of endogenous phenomena, such as the emergence of a specific EEG, EMG, oculographic or other pattern activity), combined with different shifts in activity of device-effectors, with which the user interacts within the system of biological feedback and/or brain-computer interface, and the allocation of areas of convergence of extrema in the matrix W2(a,b) describing the appearance or extinction of the chain of local maxima. Method and algorithms based on it were applied for the analysis of the background EEG and visual evoked potentials. It is demonstrated that the proposed method allows to significantly expand the understanding of the mechanisms of neural activity forming EEG: it highlights presence of several oscillators forming α rhythm, and shows the features of the dynamics of their electrogenesis. Also, there's shown a presence of components in the VEP, slowing oscillatory activity and fading within about 0.1 sec.