Connectivity analysis of EEG recordings for Epileptic patients

We propose a novel approach of connectivity analysis for EEG recordings of epileptic patients. The final goal of this study is to integrate EEG and fMRI to extract temporal and spatial information about seizure focus. However, in this paper we only report the EEG connectivity analysis. Discrete wavelet transform is applied on preprocessed EEG recordings acquired in MR scanner from epileptic patients to study the correlations between different regions of the brain in a specific frequency band during interictal epileptiform discharge (IED) and non-IED time intervals. These IED and non-IED labels are manually marked on EEG signals by an expert neurologist. The correlation matrix of concatenated wavelet coefficients of IED (non-IED) time intervals of the EEG channels are estimated. Two graphs indicating the brain functional network during IED and non-IED segments have been obtained after thresholding the IED and non-IED wavelet correlation matrices. These graphs are different for various epileptic patients. The distinct connections between these IED and non-IED graphs can provide some valuable information about the seizure focus and their inter-connections.