Independent component analysis of generalized tonic clonic seizure

Abstract

Epileptic seizures are difficult to detect and classify using electroencephalogram (EEG) due to superimposed muscle artifacts. The objective of this study is to determine features that could differentiate the abnormal EEG activity due to epileptic seizure from a normal background activity. A study group of 20 subjects suffering from a commonly occurring primary epileptic seizure, generalized tonic clonic seizure (GTCS) was compared with a control group of 20 subjects without GTCS. Independent component analysis (ICA) was used to extract independent signals from inter ictal EEG signals. Fast Fourier transform was applied to the independent components and the features were extracted. Wilcoxon rank sum test was performed to find the spectral features that could classify abnormal activity from normal activity. Mean, median, fifth percentile and power in range 2.5–4.5 Hz with P< 0.001 were the features that could differentiate abnormal activity from normal activity. Coefficient of variation, median absolute deviation, 95th percentiles were not able to differentiate normal from abnormal activity.