Diagnosis of Epilepsy Utilizing Time-Series Distribution of EEG Signals

Epilepsy is a set of chronic neurological syndromes produced by sudden and transient electrical disorders in brain. Doctors perform different experiments to detect the epilepsy and type of it. Most common and the best way to epilepsy detection is analyzing electroencephalogram (EEG) signals. In this paper, we present a new method for classifying the brain activities in order to detect epilepsy seizures. The proposed method is based on time-frequency analysis of EEG signals using stationary wavelet transform (SWT). At first, four level SWT of EEG signal is obtained and then the coefficients are down-sampled with factor 2. After that, the local binary pattern (LBP) of the obtained coefficients are calculated and LBP coefficients are modeled with GARCH model. The parameters of GARCH model construct the feature vector and K-nearest neighbor and support vector machine (SVM) classifiers are used for classification. Results show that our proposed method has the better classification accuracy than the recently proposed methods.