基于Granger因果检验和PCA的脑网络效应连接方法

为了提高脑功能效应连接网络检测的可靠性,提出了一种基于Granger因果关系检验和主成分分析(PCA)的功能磁共振(fMRI)数据效应连接方法.该方法首先通过PCA提取感兴趣区域内fMRI信号的时间主成分,以此特征作为时间参考信息,然后计算参考信息与大脑其余每个体素之间的Granger因果关系,并映射到全脑,形成Granger因果图(GCM).理论推导阐明了所提方法的有效性.采用该方法研究人脑运动功能脑区在手动任务下的效应连接GCM,结果验证了运动功能神经网络理论.

Effective Connectivity of Brain Network Based on Granger Causality and PCA

An fMRI effective connectivity analysis approach combining Granger causality (GC) and principle component analysis (PCA) is proposed, by which the causality of the selected regions in the manual movement task can be investigated simultaneously. By first using group analysis of statistic parameters mapping (SPM) to identify brain activation clusters, regions of interest (ROI) were selected. Then, temporal principal components were extracted by PCA from the fMRI signals in the ROI, and the patterns were considered as temporal reference function. Finally, Granger causality between the reference region and each of other voxels of the brain was calculated. The results were mapped into the whole brain and a Granger causality map (GCM) was obtained. The validity and reliability of the presented approach were confirmed by the theoretic and the practical fMRI results. After localizing a manual movement task-induced activation in the motor area, directional influences of physiological connectivity network between the reference and other brain regions were resolved using our new approach.