基于复杂网络的脑电信号分析

采用相空间重构方法将一维脑电信号转换成复杂网络进行研究。将一维时间序列构造成多维相空间,计算多维时间序列任两个向量点间的距离得到加权矩阵,选择恰当的阈值将加权矩阵转换成二进制矩阵,该二进制矩阵视做时间序列所对应的复杂网络的邻接矩阵,阈值的选择使得生成的复杂网络满足连通性。利用该方法对睁眼和闭眼时的脑电信号进行辨别分析。结果表明,两信号的递归图、网络拓扑图、网络度分布和模体分布均表现出显著不同。因此,此分析方法为利用复杂网络实现分析和辨识不同脑电提供了新的思路。

Electroencephalograph analysis based on complex networks

This paper investigated one dimension EEG by converting it into complex networks via phase space reconstruction. To construct a complex networks, regarded each vector point as a node, and determined the edges by the phase space distance of each pair of vector points. A selective threshold value, which made the complex networks satisfy connectivity, could transform the distance matrix into a binary matrix. The binary matrix viewed as the adjacent matrix of complex networks was used to draw network topology and to analyze network characteristics. Applied the constructing networks method to distinguish EEG during eye-open and eye-closed resting conditions. The results indicate that recurrence plot, network topology, degree distribution and motifs distribution of the two networks show a distinct difference. Therefore complex networks as a data representation framework provide a new way for analyzing and distinguishing electroencephalograph.