针对局灶性癫痫患者的脑电微状态分析

Current study of the neural mechanism of epilepsy based on electroencephalography (EEG) is limited to the epilepsy-related local brain regions. This study uses microstate to reveal the cerebral neural mechanism of epilepsy from a global perspective. The modified K-means algorithm is used to cluster typical microstates of epileptic patients, and the sequence of the microstates is obtained by fitting the microstates with the continuous EEG signals. The duration, coverage and occurrence are calculated to compare the relative trend difference of parameters between patients and healthy people. Then wavelet transform is employed to test the fractal characteristics of EEG microstates sequence. Results show that the mean duration of epileptic patients' microstate D is (101.42±22.91) ms, the proportion of microstate D in the entire time series is (30.86±9.79) %, and the occurrence of microstate D is (3.10±0.94) times per second. The values of the parameters of microstate D are significantly larger than those of microstate A and microstate B, and close to those of microstate C, which is different from the healthy people. In the time range from 32 ms to 16 s, the Hurst parameter of microstate sequence is 0.643 5±0.010 2, so the sequence possesses the fractal property. It is concluded that the anomaly of the parameters of microstate D reveals epileptic patients being prone to attention deficit, and the change of the parameters of microstate is helpful for the diagnosis of epilepsy. The fractal property of the epilepsy microstate sequence indicates that the microstate sequence has self-similarity and scale-free characteristics in a wide time range, so the microstate sequence has the potential in the pathogenesis study of epilepsy and the prediction of its seizures.