Abstract
This paper presents an automated threshold-based multi-channel epileptic seizure detection algorithm designed for low-complexity hardware implementations. The algorithm relies on two discriminative, computationally simple time-domain features, based on power and amplitude variations, that enable accurate and timely detections due to their rapid adaptiveness to fluctuations in neural activity. To ensure long-term functionality and high sensitivity, system thresholds are optimized through an offline calibration process that exploits the statistical analysis of patient-specific inter-ictal and ictal periods. The novelty of the approach lies in its multi-channel decision-making strategy, which enhances reliability against false alarms. The proposed algorithm is tested on multiple datasets to assess its adaptability to different recording conditions, achieving roughly 98% accuracy and over 98% sensitivity on both the EEG CHB-MIT dataset and the iEEG SWEC-ETHZ dataset, with average latencies of 3.37 s and 7.84 s, respectively. These results are comparable to, and in some cases outperform, several published machine-learning-based approaches. On the hardware side, FPGA synthesis highlights the minimal and scalable resource requirements of the proposed architecture, achieved through Time-Division Multiplexing (TDM) of both filtering and feature extraction. When compared to state-of-the-art proposals, the system emerges as an ideal candidate for real-time, resource-constrained hardware implementations.