Surface EMG Signal Analysis using Hand-Crafted Features for Detection and Classification of GTC seizures

Abstract

Epileptic seizures with the risk of sudden unexpected death in epilepsy affect the quality of life. Nearly, one-fourth of the individuals suffer from seizures that cannot be treated with medications. Due to the high-level possibility of injuries and complications, generalized tonic-clonic seizures have a considerable contribution to unexpected death. These generalized tonic-clonic seizures activity need to be detected and identified through brain and muscle activity, heart rates, and EMG signals. In this paper, we propose a framework for distinguishing normal activity from seizure activity along-with its categorization. Proposed framework focuses on extraction of multiple sEMG hand-crafted features with the time and frequency domain analysis. The proposed methodology for sEMG signals and for GTC class detection has been tested using multiple classifiers including KNN, SVM and ensembles. The obtained results have shown 10% improvement in classification over the state-of the-art approaches available in literature.