空间ICA在fMRI数据上的应用与分析

独立成分分析(ICA)技术试图将多维数据分解成若干个相互统计独立的分量。时间ICA和空间ICA都可以用于分析功能核磁共振成像(fMRI)数据。但由于fMRI数据空间维数远远大于时间维数，为计算方便，在分析fMRI数据时。则更多的使用空间ICA方法。本文在单任务激励实验中，利用ICA方法从fMRI数据中分离出若干个与任务相关的独立分量，其中包括与任务相关的恒定分量(CTR)和与任务相关的暂态分量(TTR)；通过将这些独立分量进行空间映射，得到了与任务相关的脑部激活区域。将此结果与SPM的分析比较，得到了一致的结果。在对结果的分析中，我们进一步指出了ICA方法的特点和局限性。

The application of spatial independent component analysis in fMRI data

Independent Component Analysis (ICA) is a statistical technique that attempts to decompose multivariate data into many statistically independent components. Temporal ICA and spatial ICA have been applied to analyze fMRI data. As there are many more spatial dimensions than temporal ones in fMRI data- spatial ICA is dominant in the fMRI analysis in order to reduce the cost of computation. In this paper- the experiment with single task was discussed. And both consistently task-related(CTR)and transient task-related(TTR) components were extracted through the ICA method. By mapping these independent components into the structural image of the brain- we got the voxels activated by the task. Comparing the results from ICA and SPM analysis- we found them almost same. During the analysis of the results- we also pointed out the ICA characters and limitations.