Title: Different approaches of analysing EEG signals for seizure detection

Authors: B. Suguna Nanthini; B. Santhi

Addresses: School of Computing, SASTRA University, Thanjavur, Tamil Nadu 613401, India ' School of Computing, SASTRA University, Thanjavur, Tamil Nadu 613401, India

Abstract: Epileptic seizures are the outcome of the transient and the sudden electrical disorder of the brain. The Electroencephalogram (EEG) is a diagnostic imaging method, which measures the electrical activity in the brain. The main principle of this study is to observe the performance of classifiers regarding Support Vector Machine (SVM) and Artificial Neural Network (ANN) using wavelet coefficients for seizure detection. This paper uses discrete wavelet transform to analyse EEG signals, which are non-stationary. The EEG signals were decomposed by db1, db2 (daubechies wavelet) and haar wavelet. Grey Level Cooccurrence Matrix (GLCM) and statistical features are extracted from the decomposed EEG signal. This work concludes that SVM classifiers using db2 with hybrid features are the best outcomes for EEG signal classification.

Keywords: EEG signals; electroencephalograms; epileptic seizures; epilepsy; GLCM; grey level cooccurrence matrix; SVM; support vector machines; ANN; artificial neural networks; discrete wavelet transforms; DWT; classification accuracy; signal classification; seizure detection.

DOI: 10.1504/IJSISE.2015.067066

International Journal of Signal and Imaging Systems Engineering, 2015 Vol.8 No.1/2, pp.28 - 38

Received: 31 Jan 2013
Accepted: 10 Nov 2013

Published online: 25 Jan 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article