An Efficient FPGA Realization of Seizure Detection from EEG Signal Using Wavelet Transform and Statistical Features

Abnormal electrical activity of a group of brain cells is known as seizure. Electroencephalogram (EEG) is most commonly used for the detection of seizure. This paper presents an efficient method for seizure detection from EEG signal using wavelet transform and statistical features in Field Programmable Gate Array (FPGA). A hardware efficient multiplierless architecture of 3/5 B-Spline wavelet transform is used for subband decomposition. Variance, standard deviation, mean energy, maximum amplitude, skewness, and kurtosis are the various features analyzed. There is a significant difference in the feature values obtained from EEG database of patient and a normal individual. The classifier used to classify the selected features is support vector machine classifier. Confusion matrix is used for measuring the accuracy of the proposed classifier. The system has two phases named training phase and testing phase. In order to train the system in the training phase EEG database of both epileptic seizure affected and normal individual is given. In the testing phase, the system will detect the presence or absence of seizure in the inputted EEG database. It is possible to fully place and route the detection system in an FPGA. The system is verified by Verilog in ISim simulator and synthesized in Virtex 6.