Feature Extraction of EEG Signal to Classify Epileptic Signal Using Neural Network

Thumbnail Image
Jazzar, Isam Mutasem
Journal Title
Journal ISSN
Volume Title
جامعة النجاح الوطنية
Electroencephalogram (EEG) is the electrical signal associated with the communication of the brain neural cells. It is used to evaluate and test the electrical activity of the brain. Consequently, it can be used to detect abnormalities associated with this activity such as epilepsy. Epilepsy, characterized by recurrent seizures, is one of the most common neurological disorder that affect people at all ages. It is associated with abnormal electrical activity in the brain. One way to detect and diagnose epilepsy is by using electroencephalogram (EEG) signal since it contains enough information to characterize the disease. We designed an algorithm capable of automate the process of identifying epileptic seizures and classifying it into three classes: normal, interictal, and ictal. The four-stage pipeline consists of a preprocessing stage, a wavelet transformation stage, a feature extraction stage, and a classification stage. The wavelet transformation stage is used to process the signals in order to prepare them for feature extraction stage. Then, statistical features are extracted from the coefficients of the wavelet transformation. Nine features were extracted and used in the classification of the signals using the Artificial Neural Network. To evaluate the performance of our model we used several measures includes: accuracy, sensitivity, and specificity. Using 300 brain signals and carrying proper calculations, we identified 144 epilepsy cases, and 156 non-epileptic cases. The accuracy, specificity, and sensitivity of our model are 81%, 80%, 84% respectively. The project provided a method to solve problems resulted from epilepsy diagnosis