Automated Epileptic Seizure Detection using Principal Component Analysis Framework

Abstract

Epilepsy is a serious chronic neurological disorder and the most common brain disease after a migraine. It has various symptoms, as loss of awareness, irregular behavior, and seizure. Electroencephalogram (EEG) waveform is a prime source commonly used for the diagnosis of epilepsy. However, the EEG-based diagnosis faces many issues in real-time situations due to the coupling of various noises and the non-stationary nature of the signals which affects the accuracy of detection of an epileptic seizure. Therefore, a machine learning approach is proposed for the automatic detection of epileptic seizures by training the model to differentiate the seizure and normal person based on EEG signal artifacts. In this work, the Principal Component Analysis (PCA) has been used to reduce the dimensionality of the EEG dataset. The results on the dataset show superior performance of the proposed approach. This approach is robust in nature and achieved the best classification accuracy of 100%. Finally, the presented method supports the high detection performance, less complexity, and the possibility to develop an automated epileptic seizure detection hardware system.