Feature Recognition And Extraction Of Epileptic Eeg

Electroencephalogram (EEG) is a well-established clinical procedure, which can provide some important information to the diagnosis of a number of brain disorders (for example, epilepsy or brain tumors). Because seizures usually occur infrequently and unpredictably, obtaining such a record might require a long-term EEG Nowadays, the tasks are mainly achieved by human vision. Visual inspection is prohibitively time-consuming and inefficient. Moreover, Visual inspection lacks standards. Automatic detection of epileptic waves can solve the problem above.The epileptic waves identification develops with the technology of modern signal processing. The main object of this paper is to propose some solutions with good performances based on the wide study on the existing methods of epileptic waves identification.In this paper, some classifier methods are studied, including wavelet transformation(WT), nonlinear energy operator(NEO), artificial neural network(ANN) and support vector machine(SVM). The major contents of this thesis are as follows:Firstly, the existing methods are studied widely. Their characteristics are analyzed.Secondly, an epileptic spike detection system that combines WT, NEO and the ANN is proposed. The system is evaluated with normal and epilepsy seizure EEG recordings. The simulation and data analysis results show that the system performs well on both normal and epilepsy seizure conditions.Thirdly, an SVM based method of epilepsy feature extraction and recognition is proposed since it is a new statistical learning theory based learning machine. The data analysis shows that the SVM is very robust for its good generalization performance in spike identification comparing with the back-propagation neural network.Finally, based on the researches above, the actuality of market in electro- encephalograph is surveyed. After analyzing characters of the new style electroencephalograph, some basic and tentative tasks are achieved on the application of epileptic waves identification in electroencephalograph.