Improved Feature Space for EEG-based Epileptic Seizure Detection Using Signal Processing Techniques

The non-stationary nature of the electroencephalogram (EEG) signal makes its analysis essential as it may point the way toward an appropriate detection technique for patients with neurological disorders, particularly epilepsy. The quality of certain variables extracted from an EEG data set that describe seizure activity is a major determinant of the effectiveness of EEG-based epileptic seizure detection. The Improved Feature Space Method (ICFS) with Discrete Wavelet Transforms (DWT) is a unique analysis technique presented in this paper for identifying epileptic seizures from EEG signals. The proposed study includes using DWT to identify the most salient characteristics from the time domain, frequency domain, and entropy-based features after first using FIR for the filtering process. After that, an ensemble of Support Vector Machine (SVM) classifiers is trained using the chosen feature set. Based on the same benchmark EEG dataset, the experimental results reveal that the suggested method performs better than the traditional correlation-based method and also surpasses several other state-of-the-art methods of epileptic seizure detection.