Classification of epilepsy using computational intelligence techniques

Epilepsy is common disorder of the brain that affects the cerebral cortex. EEG is one of the main tools for diagnosis neuron activities and brain disorders. EEG signals store important information that is useful for diagnosis neural diseases. In this work, implementation of epilepsy classifier using neural network was employed. Identification of three classes EEG signals: seizure free, pre-seizure and seizure was considered in this research, where seizure free and pre-seizure signals are difficult to identify. For comparison classification with SVM (Support method machine), KNN (K-nearest neighbour) and linear classifier has been done. Data under different noise level were tested and robustness of the classifiers checked. Best accuracy 100% and average accuracy 71% achieved in recognition three classes during the testing. Testing result shows that proposed method can be used for better classification of seizure free, pre-seizure and seizure period epilepsy.