RSeiz: A Channel Selection Based Approach for Rapid Seizure Detection in the IoMT

Epilepsy affects 1% of the world population, which necessitates a fast seizure detection system for practical epilepsy solutions. The reduction of seizure detection delay is a critical problem which needs to be addressed as rapid detection provides effective treatment. In this paper an electroencephalogram (EEG) based, patient-specific seizure detection system is proposed in the Internet of Medical Things (IoMT) framework which can detect seizures at a minimum delay. The proposed system uses neighborhood component analysis (NCA) for channel selection, statistical features for optimal feature extraction, and a ReliefFbased optimization (RBO) in conjunction with a k-nearest neighbor classifier for feature classification. A publicly available database (CHB-MIT EEG) has been used for evaluation of the proposed algorithm. The simulation results show that the proposed algorithm provides a sensitivity of 100% while maintaining a low average latency of 1.49 sec, which may be useful for practical epilepsy treatment and biomedical applications.