FPGA Implementation of High Accuracy Automatic Epileptic Seizure Detection System

Analysis of Electroencephalogram (EEG) records acquired from the brain is considered the easiest and a powerful tool in diagnosis the neurological disorders that are related to the electrical activities of the brain like epilepsy. In this paper, we propose a novel approach to extract the most relevant features to epilepsy disease and use them to do classification process that yields fast and accurate automatic diagnosis. Hilbert Transform of EEG signals is calculated and mean power frequency is applied to reduce the feature vector dimensions. The classification process between normal and seizure cases is done using Multilayer Perceptron Artificial Neural Network. The classification accuracy achieved using this method is 100 %. FPGA implementation of the proposed system is done for real-time analysis. High accuracy of the proposed system, power efficiency and its light weight make it a good candidate for automatic epileptic seizure detection.