Your browser version is outdated. We recommend that you update your browser to the latest version.

Manuscript Title: Automated Epileptic Seizure Detection inEEG Signals Using FastICA and Neural Network

Author : Sivasankari. N and Dr. K. Thanushkodi

Email : sivasankari_phd@yahoo.comthanush12@gmail.com

Abstract: Brain is one of the most vital organs of humans, controlling the coordination of human muscles and nerves. The transient and unexpected electrical disturbances of the brain results in an acute disease called Epileptic seizures. Epileptic seizures typically lead to an assortment of temporal changes in perception and behavior. A significant way for identifying and analyzing epileptic seizure activity in humans is by using Electroencephalogram (EEG) signal. In a significant number of cases, detection of the epileptic EEG signal is carried out manually by skilled professionals, who are small in number. This necessitates automated epileptic seizure detection using EEG signals. To add with, a number of researchers have presented automated computational methods for detecting epileptic seizures from EEG signals. In this article, we propose a novel and efficient approach for automatically detecting the presence of epileptic seizures in EEG signals. First, the input EEG signals are analyzed with the aid of Fast Independent Component Analysis (FastICA), a Statistical Signal Processing Technique, to obtain the components related to the detection of epileptic seizures. The BackPropagation Neural Network is trained with the obtained components for effective detection of epileptic seizures. The experimental results portray that the proposed approach efficiently detects the presence of epileptic seizure in EEG signals and also showed a reasonable accuracy in detection.

Keywords: Electroencephalogram (EEG) signals, Epilepsy, Seizures, Epileptic Seizures, Statistical Signal Processing, Fast Independent Component Analysis (FastICA), BackPropagation Neural           Network (BPNN).

Vol 1 (2)