EEG Detection Wavelet Analysis And Neural Network

When a patient has a disease of the brain or other nervous system diseases, doctors will often help of EEG as a diagnostic tool. EEG can accurately reflect many brain diseases, especially for seizures, more accurate diagnosis. EEG brain disease in patients with abnormal waveform contains a lot of these abnormal waveform mainly by spikes, spike and slow wave complexes, sharp waves, sharp and slow wave complex composition.Early clinical EEG mainly rely on manual to read the identification of abnormal EEG waveforms, this method is very time-consuming, and due to the interference artifacts, EEG physician judgment is sometimes not the same. So far there have been a method to automatically detect and EEG expert system applications,simulation method,independent component analysis,artificial neural network, wavelet analysis method.To compensate for the shortcomings of a single method is not effective, and better improve the EEG signal processing capability, this paper, the digital signal processing, wavelet analysis, genetic algorithms and neural networks combined with other methods applied to the EEG signal processing.Firstly, the digital signal processing technology,and in-depth study of the adaptive filter and variable step size algorithm.For frequency characteristics of artifacts,design variable step size adaptive filter frequency interference filter Secondly, the study of the wavelet decomposition technique using wavelet decomposition of the filtered EEG artifact frequency decomposition into different scales, the study design appropriate for different algorithms filter out artifacts artifacts characteristics.Finally, in-depth study of the neural networks and genetic algorithms,based on abnormal waves and artifacts are not exactly the same principle at different scales, the decomposition of the input EEG neural network identification. For lack of neural networks, genetic algorithms designed neural network learning process optimization.After the final test, the system can more accurately identify abnormal EEG wave characteristics, the recognition rate of87.2%.