A Low Complexity Algorithm for Epileptic Seizure Detection using Statistical Moments and Support Vector Machines
Ahmad Mohammad Sarhan
College of Engineering and Technology, Computer Engineering, American University of the Middle East, Eqaila, Kuwait.
Epilepsy or seizure disorder is one of the most common neurological disorders. Epilepsy is characterized by unforeseeable seizures. The cause of these seizures is often totally unknown. Epileptic discharges normally appear on the Electroencephalogram (EEG) signal which represents the brain’s electrical activities. EEG signal analysis is the standard approach used in the detection and prediction of epileptic seizures. A novel automated system for the identification of epilepsy patients and detection of seizures is proposed in this study. Statistical moments are used to reduce the dimension of the input EEG signal and to obtain distinctive features from it which are subsequently fed to a modified Support Vector Machine (SVM) algorithm for classification (epileptic or not epileptic). Experimental tests show that the standard deviation and mean values of the input EEG signal form robust features. Testing the performance of the proposed system on a publicly available epilepsy dataset provided by the University of Bonn, achieved 100% accuracy. The proposed system requires up to 83% fewer clock cycles than the lift algorithm and 88% fewer clock cycles than the convolution-based algorithm.
Keywords: Electroencephalogram (EEG), statistical moments, support vector machine (SVM), seizure detection, epilepsy, time complexity.
Received February 22, 2017; Revised March 29, 2017; Accepted April 16, 2017
*Correspondence: Ahmad Mohammad Sarhan, Email: firstname.lastname@example.org, Contact: +966533626554