Abnormal brain detection and analysis of EEG signals

The Electroencephalogram(EEG) is a representative signal enclosing information about the function of the brain. EEG signals can be utilized to study an automatic diagnosis of epilepsy. This later is characterized by a spontaneous and an unexpected that occurs seizures. In this paper, we propose an automatic detection of seizure disease using a support vector machine (SVM) classifier for the classification of EEG signals from three classes like healthy, pre-ictal and epileptic patients. Five statistical time domain features are extracted namely Root Mean Square (RMS), Sekwens (Sk), kurtosis (Ku), crest factor (CF) and clearance indicator (CLI). The Generalized Discriminant Analysis (GDA) is applied on statistical time domain features to reduce the dimensionality. These features with reduced dimensionality are utilized as input to a support vector machine (SVM) classifier. Our results in terms of classification accuracy out performs other existing methods and achieve the best classification rates with an average of 98.80%. Our approach is effective to diagnosis and detect epileptic seizure.