Eigenvalues of Hankel Matrix based Epilepsy Detection using EEG Signals

Epilepsy is a neurological disorder characterized by recurrent seizures which are caused by abnormal electrical activity in the brain. The electroencephalogram (EEG) is a commonly used method for detecting and analyzing seizures. Identifying subtle changes in the EEG waveform by visual inspection can be challenging. It has led to a significant domain for researchers to develop intelligent algorithms to detect such subtle changes. Additionally, the EEG signals are non-linear and non-stationary in nature which makes the interpretation and detection of normal and abnormal activity more difficult. This paper proposes an automated method for detection of epilepsy from EEG signals based on the eigenvalues of Hankel matrix. In the proposed method, EEG signals discretely segmented in time domain and each segment is represented by Hankel matrix. Eigenvalues of each Hankel matrix are extracted and considered as features for the detection of epilepsy. All the eigenvalue-based features are ranked using maximum relevanceminimum redundancy (mRMR) algorithm and optimized number of features are calculated for combination of classes in order to achieve best accuracy. Decision tree-based classifier could achieve above 99% of accuracy for classifying normal and seizure patients and and over 98% for normal, seizure-free and seizure affected patients using the proposed method. Obtained results are also compared with recent methods to justify the supremacy of the proposed method over other related methods.