Detection of Epileptic Seizure in EEG Using Sparse Representation and EMD

Abstract

Epileptic seizure detection is the most important part in the diagnosis of epilepsy. The automatic detection and classification of epileptic EEG signals has great clinical significance. This paper proposes a novel method for epileptic seizure detection using empirical mode decomposition (EMD) and sparse representation based classification (SRC). Firstly, EMD was used to decompose EEG into multiple Intrinsic Mode Function (IMF) components. Secondly, the features like variation coefficient, fluctuation index, relative energy and relative amplitude were extracted from the IMFs. Finally, in the framework of sparse representation based classification (SRC), the feature vectors of test sample were represented as a linear combination of the feature vector of training samples with sparse coefficients. Experimental results show that the time consumed by one epileptic EEG test sample is not more than 5.9 s, and the accuracy is up to 97.5%. In SRC, the raw EEG signals were replaced by extracted features, which could reduce data dimension and computational cost. The algorithm has a good performance in the recognition of ictal EEG. The higher recognition rate and fast speed make the method suit for the diagnosis of epilepsy in clinical application.