Wavelet based algorithm for the estimation of frequency flow from electroencephalogram data during epileptic seizure

Abstract

Objective: EEG data during temporal lobe seizures have been reported to show lateralized buildup of theta activity. However the exact dynamics of the theta activity and its clinical significance are not known. In this work we present an approach using wavelets to study the frequency flow dynamics of this buildup.

Methods: We employ continuous wavelet transform to obtain a time frequency representation of the EEG signal. Using a ridge extraction algorithm, the instantaneous frequency is estimated from the normalized scalogram.

Result: We found that prior to the seizure onset, frequency flow builds up to 5-12 Hz range and the duration for which the frequency remains in this range gradually increases soon after the seizure onset. We also observed buildup at the adjacent regions. Such buildup characteristics are not seen during baseline conditions of the same patients.

Conclusions: Simultaneous buildup of frequency at the temporal and the adjacent regions indicates that during seizure the neuronal interactions propagate over large regions of the brain.

Significance: Given that activity in the 5-12 Hz frequency range is seen often in the more alert state, our findings suggest that the brain might be in a transient alert state prior to the epileptic seizure.