EEG signal processing for epilepsy seizure detection using 5-level Db4 discrete wavelet transform, GA-based feature selection and ANN/SVM classifiers

Epilepsy is a neurobiological disease caused by abnormal electrical activity of the human brain. It is important to detect the epileptic seizures to help the epileptic patients. Using brain images for epilepsy diagnosis and seizure detection is time-consuming and complex process. Thus, electroencephalogram (EEG) signal analysis is focused in many papers of this field to detect the epileptic seizures. In addition, EEG signal acquisition is non-invasive and less painful for patients. However, raw EEG signal has many unrecognizable data not suitable for accurate diagnosis. Therefore, the raw EEG data must be analyzed while the features can be extracted. In this paper, discrete wavelet transform (DWT) is used to extract features of EEG signal by dividing it to five sub-bands. The proposed technique also includes genetic algorithm approach for selecting more effective features and finally, classification is performed by two strategies as artificial neural network (ANN) and support vector machine (SVM). The performance of two classifiers are compared where the simulation results show that the proposed strategy accuracy in detecting epilepsy seizures is better than other similar approaches in the literature.