DWT based epileptic seizure detection from EEG signal using k-NN classifier

This work presents a framework for epileptic seizure detection from recorded EEG signal for healthy and epileptic patient. Simulink has been used to model, EEG signal decomposition using discrete wavelet transform (DWT) and statistical calculation; implementable on FPGA with Xilinx System Generator. After DWT decomposition Mean Absolute Value (MA), Standard Deviation (SD) and Average power(AP) are extracted as statistical feature for epilepsy detection with k-Nearest Neighbor (k-NN) classifier. Results show that k-NN classifier gives better accuracy with SD and SD with MA for eyes open and epileptic seizure dataset with less number of extracted features.