Epilepsy EEG Analysis Based On Phase Synchronization

EEG is a weak non-linear and non-stationary bioelectric signals from the brain cells of the cerebral cortex that produced by the brain activities. EEG contains a large number of physiological and pathological information, so analyzing the characteristics of EEG has become the primary method to diagnose and treat the brain diseases. Thus, studying the characteristics of EEG and assessing the physiological status of the human body is significant. This paper based on the theory of phase synchronization to analyze the EEG signal from the following three aspects :First, taking advantage of the EPSI algorithm based on HHT in nonlinear and non-stationary signal, we put forward the EPSI which also based on it. We first apply the algorithm into Rossler model, the results show that it can measure the degree of synchronization of two coupled system effectively. And then, calculate the EEG EPSI which based on that algorithm, by making a comparison between normal and epilepsy EEG EPSI, we can find that the number of EEG epilepsy EPSI is higher than the normal, which show that the algorithm of EPSI based on HHT can analyze the synchronization of the EEG effectively.Second, study the APSTM of EEG about the theta, alpha, beta and gamma band, which based on the average phase synchronization time matrix(APSTM) algorithm. By comparing APSTM determinant between epilepsy and normal EEG of that four bands, the results show that APSTM algorithm of epileptic is higher than normal APSTM determinant in the corresponding of that four bands. While the theta waves determinant detects epileptic discharge more obviously when compared with its stopping discharging, the beta waves are less. But alpha and gamma waves can’t distinguish the two different period clearly. This show that APSTM algorithm can distinguish epileptic from normal EEG effectively, and the features of theta waves based on APSTM algorithm in research of epilepsy EEG are the most significant.Third, based on surrogate data test algorithm, analyze the EEG partition phase synchronization index( PPSI), which has different data length and frequency bands. First, by comparing threshold of several normal surrogate data algorithm with the associated frequency about the original signal of PPSI, found that RSS(Rank- Shuffled Surrogate) correlation coefficient is the lowest, indicating that RSS is the most appropriate algorithm. And then, EEG under different cycle length are analyzed based on the RSS algorithm, the results show that EEG signals which cycle length from about 3 to 18 is more suitable for computing PPSI, the conclusion will have positive influence for researching the coupling studies of the multi-channel EEG.