JIITA, vol.5 no.2, p.436-446, 2021, DOI: 10.22664/ISITA.2021.5.2.436
M. Sornam*, E. Panneer Selvam
Department of Computer Science, University of Madras, Chennai – 600025, India
Abstract: Epilepsy is the fourth most common neurological problem in which brain activities becomes abnormal, causing seizure or unusual behaviour and sometimes loss of awareness. An electroencephalogram (EEG) signal is used to detect problems in electrical activity of the brain that is associated with certain brain disorders. In this work, electroencephalogram signals were decomposed into the frequency sub-bands using DWT and set of statistical features were extracted from the sub-bands to represent the distribution of wavelet coefficients. Extracted features are given as an input to the neural network for classification. Later, proposed method is used to classify different sleep stages as well. Classification of EEG signal is performed using BPN and RNN network on both the seizure and sleep dataset Experimental results show that BPN best suits for seizure classification with 99.8% accuracy and RNN for sleep stage classification with 99.6% accuracy.
Keywords: Artificial Neural Network, Back Propagation Neural Network, Discrete Wavelet Transform (DWT), Electroencephalogram (EEG), Epilepsy.
Fullpaper: