Franaszczuk P J, Blinowska K J, Kowalczyk M
Biol Cybern. 1985;51(4):239-47. doi: 10.1007/BF00337149.
A parametric autoregressive model was applied to the multichannel EEG time series. Small statistical fluctuations of the spectral estimates obtained from the short data strings made possible to follow the time changes of the signals. The multiple and partial coherences were calculated for the four channel process and compared with the coherences computed between the pairs of channels. From the study it followed that the partial coherences are the proper measure of the synchronization of brain structures and their intrinsic relationships. The partial phase spectra give the information about the phase delays. The advantages of the parametric description of signals in the frequency domain in respect to the modelling of dynamic systems was pointed out.
将参数自回归模型应用于多通道脑电图时间序列。从短数据串获得的频谱估计的微小统计波动使得跟踪信号的时间变化成为可能。计算了四通道过程的多重相干性和偏相干性，并与各通道对之间计算的相干性进行了比较。从该研究中可以得出，偏相干性是衡量脑结构同步及其内在关系的合适指标。偏相位谱给出了相位延迟的信息。指出了在动态系统建模方面，频域中信号的参数描述的优点。
