EEGs Feature Extraction by Multi-Level DWT with Different Numbers of Principal Components

An electroencephalogram (EEG) is a measure used to record the electrical activity in the brain by attaching electrodes to the patient's scalp. An Epileptic seizure is a result of anomalous electrical activity in the brain, which is usually detected by EEG signals. In this work, we proposed a feature extraction method using different DWT levels and different numbers of principal components for the classification of seizure and normal EEG signals. First features are extracted from the different DWT levels the results then are used as an input to the PCA in order to reduce their dimension. As a dimensional reduction process, the different number of PCs are tested and evaluated using SVM classifier. The experimental results show that the use of SO-PCs obtains better result when it compared to the other number of PCs which reaches 100% in term accuracy and F-measure. In addition, the classification rates achieved in this research outperform several numbers of existent EEG seizure detection techniques published in the literature.