FPGA-Based System for Real-time Epileptic Seizure Detection using KNN Classifier

Epileptic seizure detection is a crucial task in the field of biomedical signal processing. Seizures are abnormal electrical discharges in the brain that can lead to various symptoms, such as convulsions, loss of consciousness, and cognitive impairment. People with epilepsy experience repeated seizures that happen suddenly and without warning. More than 70 million people worldwide suffer from epileptic seizure illness. Health professionals examine and identify the epilepsy in electroencephalography (EEG) signal using visual perception. The real time visual analysis of EEG signal is tough and challenging process. Therefore, it is essential to identify seizure event in EEG signal using a real time system in order to prevent patient from unexpected death. In this work Parks-McClellan algorithm was used to design the finite impulse response (FIR) filters, which can be used to remove the noise present in the EEG signals. The preprocessed signal is modelled using proposed weighted moving average model. From the modelled signal the features such as mobility feature, Fluctuation index and weighted spectral entropy features are extracted. The K nearest neighbor classifier is used to classify the seizure and normal EEG signal based on the extracted features. The system that has been developed employs the utilization of the Xilinx System Generator methodology on the Zynq-7000 fully programmable SOC (System-on-a-Chip) platform. This approach enables the implementation of a fully customizable and flexible system, utilizing the inherent capabilities of the SOC and the System Generator tool. Furthermore, the use of the Zynq-7000 platform enables integration of both hardware and software components, providing a more efficient and powerful solution.