A subband correlation-based method for the automatic detection of epilepsy and seizure in the dual tree complex wavelet transform domain

In this paper, a sub-band correlation-based method is proposed for the automatic detection of epilepsy and seizure. The analysis is carried out by decomposing the electroencephalogram (EEG) signals, collected from a publicly available EEG database, into the dual tree complex wavelet transform(DT-CWT) domain. An Artificial Neural Network(ANN) is employed as a classifier where the maximum cross-correlation among the DT-CWT sub-bands are utilized as the features. Studies are conducted using EEG signals for four clinically relevant classification cases which include healthy vs seizure, non-seizure vs seizure, ictal vs inter-ictal and finally, healthy vs inter-ictal vs ictal recordings. The ANN-based proposed method provides 100% accuracy with 100% sensitivity and 100% specificity for the first three cases and also a high accuracy for the fourth case. In addition, the proposed method is computationally fast in comparison to the several time-frequency and EMD-based algorithms available in the EEG literature.