Research On Phase Synchronization Characteristic For Electroencephalograph

Brain science studies brain and intelligence mechanism from different levels and the coverage of the subject is very wide. In this thesis a series of work was launched focus on phase synchronization analysis for electroencephalograph (EEG). The analysis methods were applied to different application fields and improved based on specific application environments and detailed data characteristics.Firstly, a traditional phase locking value calculating method which includes band-pass filtering, instantaneous phase calculating and phase locking value calculating was proposed. This method could effectively separate the phase component from amplitude component and excavate the important information hidden there. Phase synchronization analysis method based on phase locking value was used to analyze the output data of an improved olfactory model which combines KⅢ model and Liljenstrom model. Simulated experiments results show that the final output of improved olfactory model reflects abundant synchronization characteristics and it could effectively reveal the change of sensitivity degree for improved model with the change of gain parameter.Traditional phase locking value calculating method has obtained great experimental effects on analysis of improved olfactory model output. Then, a set of real hypoxia EEG data was analyzed by the same method. According to the characteristics of EEG, an analysis strategy including common average reference (CAR) spatial filtering and traditional phase locking value calculating was proposed. Using this strategy on hypoxia EEG, experimental results show that phase synchronization extent among brain zones represents linear relationship with hypoxia extent. Phase synchronization extent increases while oxygen concentration becoming higher. Phase synchronization extent decreases while oxygen concentration becoming lower.Aiming to the problems existing in traditional phase locking value calculating method, an improved method based on Hilbert-Huang transform was proposed. On this basis, EEG analysis strategy was optimized further. In the new strategy, EEG data is preprocessed by both ocular artifact elimination and spatial filtering. Meanwhile, phase locking values are calculated by the improved method. Applying the new strategy on hypoxia EEG, experimental results show that the improved method has good performance on hypoxia EEG data and the precision rate is much higher than that of the traditional calculating method.