Index for Assessment of EEG Signal in Ischemic Stroke Patients

Abstract

Stroke is an acute condition of sudden compromise in vascular perfusion of brain and manifestation of neurological deficit. Worldwide, stroke is the second leading cause of death and also the third leading cause of morbidity and disability. Electroencephalogram (EEG) is a noninvasive method that captures the electrical activity of brain as a signal from the scalp. Detection and reduction of artifacts play an important task to acquire clean EEG signals so as to examine and detect brain activities. In this work, EEG signals from normal and subjects with acute ischemic stroke (AIS) are acquired under standard signal acquisition protocol from public database. The quality of the signal is improved by the techniques. An attempt has been made to detach artifacts by independent component analysis. The preprocessed EEG signal is decomposed by discrete wavelet transform method into wavelet coefficients to reduce the signal dimension. The decomposed signal is categorized as the sub-waves namely alpha, beta, delta, theta and gamma. The index such as delta–alpha ratio (DAR), delta–theta to alpha–beta ratio (DTABR), brain symmetry index (BSI) obtained by Welch's method helps to distinguish AIS from controlled subject. Also, the performance of the procedure is evaluated by statistical measures such as skewness, kurtosis, entropy, mean and variance. It is observed from the results that AIS patients have a high DAR, DTABR and BSI. Results also demonstrate that the extracted statistical metrics are high for AIS compared to that of normal individuals. Thus, the index and statistical metrics used in this work are significant in classifying AIS from normal subjects.