Epileptic EEG Signal Classification using Exponential Energy and SVM

Electroencephalography is an important method for the diagnosis and analysis of epilepsy. Epileptic seizure indicates severe brain disorder which occurs because of abrupt movement of neurons in brain. A number of methods are proposed to distinguish epileptic patients. Here the exponential energy of the time-series Electroencephalogram signal is used as an important parameter for classification of epileptic EEG. Support vector machine model is used to distinguish patients with epilepsy. We have considered records A-E, B-E and D-E for analysis and achieved an average accuracy of 87.77% in identifying epileptic Electroencephalogram without segmenting the records. An average accuracy of 100% is obtained with segmented EEG records.