In resting-state, the low frequency (0.01 - 0.08 Hz) fluctuation (LFF) of blood oxygenation level dependent (BOLD) signal has been assumed to reflect spontaneous neuronal activities. Studying the resting-state BOLD signal helps understand the activities in the brain. Studying the resting-state brain function of normal adult subject will be helpful in studying the task-state and resting-state brain function of patients. Selecting regions of interest (ROI) is an important step in functional magnetic resonance imaging (fMRI). A correctly selected ROI will do great help in fMRI analysis, while an incorrectly selected ROI will lead to wrong results. The aims of this study are to invest how to appropriately select ROIs in resting-state for further use by functional magnetic resonance imaging, and to perform a brain network analysis with these results to enlarge our understanding of resting-state brain function. Regional homogeneity (ReHo) is a new fMRI data analysis method. Based on the work, we first used the ReHo approach to obtain a functional brain in resting-state human brain. Then, a statistical analysis was performed on the ReHo maps and a group-level functional map was obtained. This group-level functional map was more suitable for functional analysis of the corresponding group. Based on this result, we proposed to perform watershed segmentation on this functional map to isolate the connected functional clusters. Compare to traditional ROI selection based on the structural template, ROI selected by template always tends to be symmetrical, and larger regions won’t be able to deal with the differences in the region. Functional clusters we obtained based on the ReHo approach resolved these problems. Brain function is organized under two basic major organizational priciples: functional segregation and functional integration. ReHo based functional segmentation concerned on the functional segregation while network analysis based on these functional regions concerned on the functional integration. We selected four ROIs in the segmentation result and performed a correlation analysis which found four frequently reported networks. This also validated the functional segmentation method. We also found that most of the brain regions were organized into a correlated and its anti-correlated network.
修改评论