EEG Signal Correlation Analysis Based On AR-Copula Model

In the field of modern medical diagnosis especially for diagnosis of brain disease, EEG is playing a more and more important role. The characteristics of many diseases will change the status of the brain with the change of time. EEG also has a corresponding change. Through the characteristics of the changes of EEG in early disease of generation, it is a great help for the diagnosis and control will be also bring the effective treatment of diseases.The article here mainly discusses the correlation problem in the EEG signal processing. Under the respect of the correlation, the article also considers the feasibility of the application of the Copula theory on the EEG signal analysis. The Copula-based correlation method is used in the preprocessing of data in the analysis of EEG signals and the diagnosis in the diagnosis of epileptic foci.(1)In the preprocessing stage, AR-Copula based tail dependence is introduced to analyze the correlation between the EEG signal and the noise signal, which will be useful in the noise detection. That will help us to use the Fast-ICA method to separate the noise from the source signal in time. This method could detect the contaminated data section affected by eye blink automatically, and so that it could decrease the iteration time of the Fast-ICA method.(2) In the stage of signal processing, diagnosis of epileptic foci usually rely on clinicians according to their own experience with EEG signal graph corresponding to analysis. For the characteristic of the EEG signal, the Copula based Kendall correlation is used in the analysis of the EEG signal, the measurement of the tail dependence is used to detect the spike wave in the EEG signal which is the standard waveform of the epilepsy. The method help to reduce the burden of the clinician in the diagnosis of the epilepsy and enhance the veracity of the EEG diagnose.For the complex of the EEG and the difficult in EEG analysis, the issue gives the application of the noise detected and eliminated based on the AR-Copula and the application of the Copula-based Kendall correlation in the diagnosis of epileptic foci. The presented methods can be the assistant tool for real the clinical applications.