Feature Extraction And Recognition Of Epilesy Signals Based On SVM-HMM Mixture Model

Epilepsy is caused by the sudden brain neurons abnormal discharge, make the nerve center produces a temporary functional disorder repeatability onset of chronic diseases. For a repetitive episodes of this is a hindrance, in the clinical medicine through 24 hours real-time monitoring in patients with brain electric signal, the doctor and by observing the eeg signals by experience whether epileptic seizures. In academic theory, epilepsy is also as a research hotspot of EEG signals, but a lot of epileptic EEG signals processing methods were focused on how to obtain a more accurate classification result, and for quantitative differences of epileptic EEG and normal EEG signals and physical meaning but not very good explanation, so cannot be combined with clinical medical practical experience to analyze the signal processing result.In this paper, the main work has the following several aspects:1) first for epileptic EEG feature extraction is optional sex problem, in this paper, based on wavelet transform coefficient of the scale of the meaning of degree of original signal is similar to the original signal, wavelet transform to extract the wavelet coefficient of the spine slow composite wave in epileptic hair EEG signals under different frequency scale, feature extraction, this process is a combination of epilepsy with clinical experience, solve the randomness of feature extraction.2) secondly, in view of epileptic EEG signals in the clinical diagnosis of empirical quantitative coding, similar degree through the SVM classification, and the classification results of each channel to recode, and finally by the HMM for new code judge brain electrical signal timing state of modal and EEG signals are classified, by the result of the HMM classification can reverse track abnormal signal frequency source, through the analysis of the sources of seizures during different frequency band when we can get a seizure of accurate quantitative analysis.3) finally, in view of the previous algorithms in the classification of epileptic EEG signals is retrospective, classification results and clinical experiences to introduce the results of the analysis of problems, this paper combined with clinical experience of the knowledge and the proposed algorithm combining the classification results, through reverse analysis for clinical experience to find reliable theoretical basis and rich experience of clinical diagnosis of epilepsy hair, solve the previous methods of epileptic EEG signals classification results is retrospective, and epilepsy with clinical experience to introduce the results of the analysis of problems.