FPGA-based Neural Network System for Epileptic Seizure Detection

An epileptic seizure is the maximum replied persistent neurological disorder. The non-stop tracking and early detection can reduce the effect of seizures. In this context, we aim to realize a hardware implementation system based on a multi-layer perceptron able to detect epileptic seizure activities. The designed system has been implemented on FPGA. We have proposed an optimized medical diagnostic approach based on wavelet transform with a minimum number of features, able to correctly classify the appropriate Electroencephalogram (EEG) classes. The optimized approach leads to minimizing the size of the FPGA prototype and reducing power consumption. The proposal system performance has been realized in terms of precision rate and the number of resources used in hardware implementation.