Method and software for processing daily EEG signals for detection of epileptic seizures in humans

Abstract

A method, an algorithm and a software tool for processing daily EEG signals for computer electroencephalographic systems to detect the manifestation of epileptic seizures in humans have been developed. Mathematically, the daily EEG signal is presented as a random sequence of white Gaussian noise zones and additive mixtures of different-frequency harmonic components. Harmonic functions interpret the manifestations of epileptic seizures. The core of the method of processing daily EEG signals is a time-shifted window inter-covariance processing with multiple kernels in the form of different-frequency harmonic functions. Based on the method of window processing, an algorithm and a software tool for daily EEG signal processing with a graphical user interface using the MATLAB environment have been implemented. The developed software can be used as a component of computer EEG systems. The results of daily EEG signal processing using the software are displayed in the form of averaged products of covariance results (the value is measured in power units) within each processing window, which quantitatively reflect the time points of epileptic seizures in a person. Manifestations of epileptic seizures are reflected through the increase in the averaged values of the power of covariances in relation to observation intervals without corresponding manifestations of these seizures. To ensure the authorization of the process of determining the level of decision-making regarding the moments of epileptic seizures (exceeding the normal level), the threshold algorithm and the Neumann-Pearson statistical criterion were applied.