Robustness Analysis Of Correlation Methods Based For Eeg Signal

The synchronization plays an important role in the information transmission and process between the different areas of a brain. The abnormal synchronization between the EEG series is also an important factor to reveal the disorder of brain. The analysis of synchronization between the EEG series is the key for describing the correlation of each area in the brain. In the thesis, we make use of some different models and the real data of EEG series to test the existing methods. The robustness of these methods is evaluated by the different contaminated data with noise, the aim is to find which method is better to treat with the different EEG signals.The simulation analysis includes dynamics system model, chaotic models, and the nerve group model of coupled two-area. The effect of the different correlation strengths and coupling strengths between two systems is considered to test the robustness of different method.As for real EEG signals, considering the effects of various factors there are numerous disturbing signals derived from recording data in the EEGs. The ICA (Independent Component Analysis) may decompose the mixed EEG series. Then we use a harmonic wavelet transform to extract the alpha wave from EEG series, and analyze it by a surrogate data. The reason for analysis the alpha wave is its significant importance in the cognitive process.The results of simulation analysis of the robustness based on models and real EEGs data indicated that the robustness of these methods is related to the different models and data. However, we can find the best one from all the methods for a certain model.