Epileptic seizure characterization using transform domain

Abstract

In case of epileptic seizure, the subject usually maintains silence. Sometimes it cannot be identified whether the subject is silent or has been attacked by epilepsy. In this paper, authors have taken an attempt to distinguish among these two cases. The approach is very simple and does not make any complicacy by introducing the classifiers. It is based on the efficient features only. Initially, the spectral features have been obtained using Fourier transform and chirp Z-transform. Further for characterizing the specific frequency, the time-frequency analysis using wavelet transform has been done with the cepstral coefficients depicted in the result. The cepstrum of the said EEG signals shows clearly the difference among them. Fast Fourier transform (FFT), chirp Z-transform (CZT) and discrete wavelet transform (DWT) have been exploited to characterize the signals.