Classification of epileptic and non-epileptic EEG events by feature selection f-score

Epilepsy is defined as a collection of symptoms and clinical signs are emerging due to intermittent brain dysfunction, which occur due to loose or excessive abnormal electrical discharges of neurons in paroxysmal with various etiologies. In this article the implemented software detection of disease epilepsy, characteristics which will represent in the detection of epilepsy and not epilepsy are from 19 electrodes, FP1, FP2, F7, F3, Fz, F4, F8, C3, Cz, T3, T4, T5, T6, P3, P4, Pz, O1, O2. The signals extracted based on the statistical characteristics of the mean, variance, standard deviation, skewness, kurtosis, minimum, maximal, correlation, energy, each electrode will produce 9 features, feature ability to detect epilepsy and non-epilepsy were analyzed using feature selection methods f-score best feature selection results will be tested using a classification algorithm Backpropagation Neural Network (BPNN). The results of 5-fold cross validation) shows that the characteristic feature vector originated from standard deviations from the electrode P4, Cz, FP1, Pz, T3, O2, C4, C3, P3, T5, O1, F2, FP2, F4, F3, T6, F8 the Maximal feature vector is Pz, Cz, FP1, P4, minimum feature vector is P4, FP2, FP1, Cz, C3, Pz, P3, F3 can detect epilepsy compared to the other electrode with a mean level of 96.2% accuracy.