基于脑电信号因效性脑网络的情感状态分析

Analysis of emotion states based on electroencephalogram signal effective brain network

提出一种基于脑电信号因效性脑网络的情感状态分析方法。采用基于向量自回归模型的Granger因果关系,计算多通道脑电信号的因果连接矩阵,并通过阈值化处理获得邻接矩阵,构建因效性脑网络。从脑网络的信息流和邻接矩阵的拓扑属性两个方面,对比分析积极和消极情感状态的全脑区因效性脑网络的差异性。实验结果表明,消极状态大脑全脑区的信息流分布更加密集、信息交互更加频繁;积极和消极情感状态时期,大脑的额叶、顶叶和颞叶等3个脑区均有较高地拓扑属性值,在情感产生过程中活跃程度更高,与情感活动紧密相关。

An affective state analysis method based on the brain network of electroencephalograph(EEG)is proposed.Based on the Granger causality of vector auto-regressive model,the causal connection matrix of multichannel EEG signals is calculated,and the adjacency matrix is obtained by thresholding,and then the effective connectivity brain network is constructed.Experimental results show that the distribution of information flow in the whole brain area of the negative emotion state is more intensive and the information interaction is more frequent.The three brain areas of frontal lobe,parietal lobe and temporal lobe have higher topological property values during the positive and negative emotion state,and they are more active in the process of emotion production,which is closely related to emotion activities.