Feature extraction and classification of EEG for automatic seizure detection

One of the many challenges in the automated detection of epileptic seizures is to draw a line of demarcation between seizure activity and non-seizure activity. To accomplish this task, identification of related features and there extraction from the EEG plays a key role. The work presented in this paper is a part of an overall effort going on to develop a new method for automated detection of seizures. A wavelet based feature extraction technique has been adopted. Statistical features, Inter-quartile range (IQR) and Median Absolute Deviation (MAD) also form part of the feature vector. The algorithm was evaluated on 23 subjects with 195 seizures. The results gave an average detection accuracy of 96.5%. The database used is the CHB-MIT scalp EEG database. All the calculations were performed on Matlab.