Temporal Characteristics of Wavelet Subbands of Epileptic Scalp EEG Data Based on the Number of Local Min–Max

Abstract

Epilepsy is a chronic brain disorder characterized by recurrent seizures. An electroencephalogram (EEG) which records the electrical activity of the brain can help diagnose seizures. Temporal characteristics of the EEG provide an insight into the states of the brain including epileptic seizures. In this study, the temporal characteristics of epileptic scalp EEG subband signals obtained using the discrete wavelet transform associated with various states of the brain including the pre-ictal, ictal and post-ictal states, are examined using a simple computational measure, referred to as the number of local min–max. From the computational results, it is observed that in any wavelet subband the EEG subband signals associated with different states of the brain exhibit distinguishing characteristics of the number of local min–max. The most remarkable temporal characteristics of EEG subband signals can be observed in the $D_1$ and $A_3$ subbands which, respectively, correspond to the highest and lowest frequency components of the EEG signals. In particular, during an epileptic seizure activity the computational results suggest that there is an increase of amplitude regularity of the highest frequency components while there is a decrease of amplitude regularity of the lowest frequency components. Furthermore, the computational results show that the number of local min–max of the $D_1$ and $A_3$ subband signals of epileptic EEG can be potentially useful for epileptic seizure classification and detection accompanied with further digital signal processing and analysis.