Non-invasive Methods for EEG Spectrum Analysis
Tatiana R. Bogatenko1, Konstantin S. Sergeev1, Galina I. Strelkova1, J. Kurths2;
1Saratov State University, Saratov, Russia; 2Potsdam Institute for Climate Impact Research, Potsdam, Germany
Abstract
In this research we propose a novel approach for classifying different stages of anaesthesia in rats using a machine learning method. Two-channel EEG data from two groups of laboratory rats was analysed. One group experienced the influence of gas (isoflurane) anaesthesia, while the other underwent the influence of drug anaesthesia. Each rat experienced three stages of anaesthesia successively: light anaesthesia, deep anaesthesia and lethal anaesthesia. The task was to determine the degree of anaesthesia non-invasively using the spectral characteristics of the EEG signals. We show that it is possible to successfully cluster the data in accordance with the anaesthesia degree using the K-Means algorithm and determine the time stamps of anaesthesia degree change.
Because the abovementioned method allows to identify spectral changes in the signals, we additionally analyse the behaviour of separate brain rhythms of the animals under different stages of anaesthesia. Using the spectral energies of five brain rhythms (delta, theta, alpha, beta and gamma), we show that such characteristics of the rhythms as mean values and standard deviation change with the changes in anaesthesia degree. We also highlight the fact that different brain rhythms prevail in different anaesthesia stages in terms of the value of their spectral energy.
Speaker
Tatiana R. Bogatenko
Saratov State University
Russia
Discussion
Ask question
