Wavelet entropy based EEG analysis for seizure detection

Abstract

Electroencephalograms (EEGs) are records of brain electrical activity along the scalp which contain precious information related to the different state of the brain. Epilepsy is the most prevalent neurological disorder in the human among dementia, traumatic brain injuries, stroke and others. Electroencephalogram being the non-stationary signals, proper analysis is highly required to categorize the normal, interictal and ictal EEGs. In the present work wavelet entropy based feature is used to differentiate the normal EEGs of healthy subjects, interictal and ictal EEGs of epileptic patients. The EEG signals are decomposed into different sub-bands using discrete wavelet transform (DWT) to obtain the detail and approximation wavelet coefficients. These coefficients are used to take out the quantitative value of wavelet entropy feature. Wavelet entropy values of different data sets are used for analysis of the epileptic EEG through statistical analysis. The t-test statistical method is used to discriminate between healthy subject (with eyes open) and ictal with 99.9% discrimination probability and interictal and ictal with 100% discrimination probability.