ANALYSIS OF EEG SIGNALS AT DIFFERENT

Abstract

Electroencephalogram signals commonly known as EEG is a representative signal containing information about the condition of the brain. Currently analysis of electroencephalogram (EEG) remains a challenge owing to limited understanding of the signal origin which leads to the complication of developing evaluation techniques. Despite of these limitations, EEG is still a valuable reflection of many physiological factors modulating the brain. It is a very important tool in the evaluation of some neurological disorders. A number of proposed methods to quantify the information content of the EEG have established. Some of these methods include Fourier Transform, entropy and Wavelet transform. This project is an attempt to analyze EEG signals during different mental conditions such as Normal, Epileptic and Alcoholic by using AR Modelling technique. As human observer cannot directly monitor the information about the state of the human brain therefore the EEG signal parameters, extracted and analyzed using computers are highly useful in diagnose. The properties of measured EEG are computed by applying PSD (Power Spectra Density) estimation for selected representative EEG samples. This project deals with a comparative study of the PSD achieved from normal, epileptic and alcoholic EEG signals. The power density spectral was estimated using Burg’s method, one of the Auto Aggressive Modelling techniques. With the help of EEG analysis software the data achieved are then extracted and fed to GMM Classifier. Different sets of train and test data were selected to train and test the Classifiers in order to achieve the best Classification rate. The results for these 3 different classes of EEG signals are tabulated.