Early detection of epileptic seizures by entropy-based methods

This work presents a novel method for early detection of epileptic seizures from EEG data. Seizure detection was accomplished in three stages: multiresolution overcomplete decomposition by the a-trous algorithm, feature extraction by computing power spectral density and sample entropy values of sub-bands and detection by using z-test and support vector machine (SVM). Results highlight large differences between the subband sample entropy values for the epileptic and the control subjects, respectively, reveling a substantial increase of such parameter during the crisis. This enables high detection accuracy and specificity especially in beta and gamma bands (16-125 Hz). The detection performance of the proposed method was evaluated based on the ground truth provided by the expert neurophysiologist as well as by objective indexes when two crisis had been recorded.