目的探讨静息状态功能磁共振数据预处理步骤中去除与非去除全脑平均信号对功能网络连接性的影响。材料与方法23例健康受试者接受静息状态功能磁共振成像；以后扣带回(posterior cingulate cortex，PCC)及前额叶背外侧区(dorsal lateral prefrontal cortex，dLPFC)分别为种子区，用时程相关分析方法分别获取默认网络及执行网络以及其负相关网络；在其他图像预处理条件一致的情况下，计算去除与非去除全脑平均信号时功能网络连接性在组内特征，及配对双样本t检验的统计学差异。结果以PCC、dLPFC为种子区的组内分析显示能定位默认网络及执行网络，并能获得其负相关网络；经Alphasim校正的P＜0.005统计阈值下的配对t检验发现，非去除较去除全脑平均信号的PCC、dLPFC正相关网络连接性在全脑水平显著增强。结论本研究证明了去除与非去除全脑平均信号两种模式预处理产生不同的脑网络特征，非去除全脑平均信号导致全脑功能连接性向正相关偏移。所以在体素水平的脑网络分析中，全脑平均信号的预处理问题应给予足够重视，建议在国内的相关磁共振脑功能连接分析研究中同时展示去除与非去除全脑平均信号的结果。%Objective:To examine the effect of global signal correction on the functional connectivity in resting-state magnetic resonance imaging network analysis.Materials and Methods:Twenty-three healthy subjects underwent 5 minutes resting-state BOLD fMRI scanning. A method of time-series correlation analysis based on seed regions of posterior cingulate cortex (PCC) and left dorsal lateral prefrontal cortex (dLPFC) was employed to extract default mode network and executive control network. The resulting functional connectivity was compared between with and without global signal correction, while the other preprocessing conditions were identical.Results:The seeds correlation methods could extract default mode network and executive control network which coincided with previous studies, and could also acquire their negative correlated networks. Paired two-samplet test underP<0.005 statistic level with alphasim correction indicated that there were significant increased positive correlations at almost whole brain level.Conclusion:These results from this study show characteristic differences between the networks with and without global signal correction. The correlation coefifcients were positively biased in the methods without the whole brain signal correction, so more attention should be paid to the issue of global signal.
展开▼
