静息态功能磁共振数据处理方法的应用

Application of the methods in analysis of resting-state fMRI data

目的比较静息态fMRI数据处理方法中的分数低频振幅(fALFF)、局部一致性(Reho)和功能连接(FC)在数据分析中的侧重点和敏感性。方法采集10名正常志愿者和10例脑干梗死患者亚急性期静息态fMRI数据,采用各种算法分别处理数据,提取患者的静息态脑网络(RSN),并比较其相对正常人改变的区域。结果 fALFF和Reho提取的患者RSN包含脑默认网络后部和初级视皮质,FC提取的RSN与脑默认网络基本一致。比较患者与正常对照,Reho方法无有意义结果,fALFF和FC得到不同结果,选择不同ROI的FC也得到不同的结果。结论 fALFF与Reho侧重脑局部功能活动的分析,FC侧重脑整体功能活动的分析;Reho对本数据敏感性最差,而FC最好,其中ROI的选择对结果影响很大,依据本数据的特性选择比依据先验知识选择得到的结果敏感性更高。

Objective To compare the specificity and sensitivity of fractional amplitude of low-frequency fluctuation （fALFF）, regional homogeneity （Reho） and functional connectivity （FC） in resting-state fMRI data analysis. Methods Resting-state fMRI were acquired in 10 subacute pontine infarction patients and 10 normal controls. The data were analyzed with different methods. The activations of resting-state networks （RSN） of the patients were taken and compared with nor- mal controls, and then the increased and decreased brain regions were demonstrated. Results In fALFF and Reho, the RSN of patients mainly distributed in posterior default mode network （pDMN） and primary visual cortex. In FC, the RSN mainly distributed in DMN. Compared with normal controls, there was no effect in Reho, and there were different results in fALFF and FC. The results in FC were influenced by different ROI. Conclusion fALFF and Reho are methods for local activity （segregation）, FC are for integration of distinct brain regions. For these data, the sensitivity of methods from bad to good is Reho, fALFF, FC. ROI chosen in FC has significant impact on the sensitivity during group analysis, based on the functional activation maps from it＇s own data is better than previous literature.