The Recognition Algorithms Of Epileptic EEG And Their Applications

Electroencephalogram (EEG) is an effective method to detect abnormal symptom of brain in clinical medicine, it can provide doctors reliable information for abnormality detection. Epilepsy, the representative of brain disorders, is so paroxysmal that long term EEG monitoring is required in analyzing and characterizing seizures and other epileptic activity. But it is very time-consuming and subjective if all this work is accomplished by person. As a result, automatic detection and classification of seizure is essential in long term EEG monitoring.At present, numbers of signals processing and analyzing method are used in EEG research work and have good performance. After looking through all of them, this paper proposes several new algorithms based on Support Vector Machine (SVM), wavelet and Empirical Mode Decomposition (EMD) for EEG analysis and seizure detection and experimental results show that these algorithms can divide seizures from normal EEG effectively. The rest of the paper is organized as follows.First, in this paper, we describ the mechanism and physiological character of the EEG waves, study the theory and method on extraction and analysis of EEG features, and review the history and the development of various methods as well as their advantages and disadvantages. All of them are the foundation of the latter research.Second, Review main algorithms in seizure detection and make a comparison of them, after studying support vector machine(SVM), wavelet analysis and empirical mode decomposition(EMD) in detail, a new algorithm is proposed in EEG signals classification. The experiment results show that the proposed method performs good in seizure detection, the recognition rate is up to 91% or more, the maximum one is 100%.Finally, this paper makes an evaluation on the prospect and application of EEG classification and epilepsy feature extraction.