The Feature Recognition For The Epileptic Waves Based On EEG Data By Using The Complexity

Epilepsy is a serious disease with the transient abnormality of the central nerve system, which is from the super-excitability and the synchronous firing of neurons. Electroencephalogram is a well-established clinical procedure, which can provide some important information to the diagnosis of epilepsy. However, a long-term EEG is often needed for the clinical doctors to view to make it time-consuming and to be lack of the standards. The automatic detection of the epileptic waves can be helpful to 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 the epileptic waves identification.In this paper,approximate entropy (ApEn) theory and its property are studied,and the clinical epileptic EEGs were analyzed by the approximate entropy,it was demonstrated that the epileptic discharges could be detected from the background EEG by the approximate entropy. In the term of the limitations of the approximate entropy in the epilepsy detection, this thesis analyzes the epileptic EEG signals with the sample entropy (SampEn) approach,to supply much higher precision than the approximate entropy. The results show that although both approximate entropy and the sample entropy decrease significantly while the epilepsy, the sample entropy is more sensitive to the EEG changes caused by the epilepsy.Futhermore, the complexity method combining with the wavelet transform is proposed to achieve the feature recognition of the epileptic waves, because the sample entropy is powerless in all cases of the epilepsy. It is proved by the project that this method can consume less computation source and is simple to be effectively carried out.