Automatic detection of absence seizures with compressive sensing EEG

Absence epilepsy, a neurological disorder, is characterized by the recurrence of seizures, which have serious impact on the sufferers' daily life. The seizure detection has a great importance in the diagnosis and therapy of epileptic patients. Visual inspection of the electroencephalogram (EEG) signals for detection of interictal, pre-ictal and ictal activities is a strenuous and time-consuming task due to the huge volumes of EEG segments that have to be studied. In this study, we proposed a novel automatic detection method based on the altered compressibility of EEG during the three states with compressive sensing. To evaluate the proposed method, segments of interictal, pre-ictal and ictal EEG segments (100 segments in each state) were used. Two entropies, namely the Sample Entropy (SE) and the permutation Entropy (PE), and Hurst Index (HI) were extracted respectively from the segments to compare with the proposed method. Significant features were selected using the ANOVA test. After evaluating the performance of the selected features by four classifiers (Decision Tree, K-Nearest Neighbor, Discriminant Analysis, Support Vector Machine) respectively, the results show that the proposed method can achieve the highest accuracy of 76.7%, which is higher than HI (55.3%), sample entropy (71%), and permutation entropy (73%). Hence, the altered compressibility of EEG with CS can act as a good biomarker for distinguish seizure-free, per-seizure and seizure state. In addition, compressive sensing requires less energy but has competitive compression ratio compared to traditional compression techniques, which enables our method to tele-monitoring of epilepsy patients using wireless body-area networks in personalized medicine.