A Novel Method for Epileptic Seizure Detection Using Separable Gabor Wavelets

In clinical therapy, epileptic seizure detection is receiving more and more attention. The analysis of scalp electroencephalograms (EEGs) is a widely used method for detecting brain abnormalities related to seizure onset. This paper presents a novel technique for the classification of epileptic seizure based on low-complex separable Gabor wavelet. Initially, the continuous wavelet transform is used to create an EEG scalogram sequence, which represents the time-frequency data. Next, the distinct features from the EEG sclogram are extracted using separable Gabor wavelet filter bank which requires low computational complexity. For the purpose of classifying seizures, the several features are concatenated into a fixed-length feature vector. The linear discriminant analysis approach is then used to reduce the dimensionality of the retrieved features significantly. Lastly, to distinguish between seizure-free and epileptic seizures, the most discriminative variables are given into a support vector machine (SVM) classifier. The proposed method’s classification performance is evaluated against a number of cutting-edge epileptic seizure detection methods. The average accuracy of the proposed method for all categories of seizure detection is found to be 99.10% with average sensitivity and specificity of 99.02% and 99.18% respectively. The results of the experiments performed on a real EEG dataset show that the proposed technique can successfully classify an epileptic seizure by extracting relevant context information from various angles.