Functional And Anatomical Connectivity-Based Parcellation Of Human Cingulate Cortex

ObjectiveIn this study, we used resting-state functional MRI and diffusion tensor imaging, to parcellate human cingulate cortex(CC) based on resting-state functional connectivity(rsFC) patterns and anatomical connectivity(AC) patterns, to analyze the rsFC patterns and the AC patterns of different subregions, and to recognize if the parcellation results obtained by two different methods were consistent.Subjects and methodsForty-seven healthy, right-handed subjects were enrolled(29 males; mean age:26.3 years, range: 20-40 years). Structural MRI(T1WI), resting-state functional MRI(rs-fMRI) data and diffusion tensor imaging(DTI) were acquired. All MR images were acquired using a GE Discovery MR750 3.0 Tesla MR scanner with an eight-channel phased-array head coil. Rs-fMRI and DTI data were acquired by a single-shot echo planar imaging(EPI) sequence. During rs-fMRI scans, all subjects were instructed to keep their eyes closed, to stay as motionless as possible, to think of nothing in particular, and not to fall asleep.Region of interest(ROI) of the CC was delineated manually in the Montreal Neurological Institute(MNI) space. The T1 WI, rs-fMRI and DTI data were both preprocessed. After preprocessing, the T1 WI was firstly co-registered to the mean functional image in native rs-fMRI space, and then transformed to the MNI space.Finally, the inverted transformation parameters were used to transform the ROI from MNI space to the native rs-fMRI space. Similarly, the T1 WI was firstly co-registered to the B0 image in native diffusion space, and then transformed to the MNI space, the inverted transformation parameters were used to transform the ROI from MNI space to the native diffusion space. Finally,the ROI of every subjects in native rs-fMRI space and diffusion space were achieved. For each individual fMRI dataset, Pearson correlation coefficients between the time series of each voxel within the ROI and that of each voxel of the whole brain were computed. For each individual DTI dataset, we estimated the fiber connections between each voxel within the ROI and every voxel of the whole brain. Cross-correlation matrix of each ROI voxel was calculated andfed into a K-means clustering algorithm for automatic parcellation. Cross-validation was used to determine the number of clusters. Then the maximum probability map for each CC subregion was calculated. Finally, functional and anatomical subregions model of cingulate cortex were established. We further analyzed the rs FC pattern of each functional subregion and AC patterns of each anatomical subregion to verify different subregions with heterogeneity. Meanwhile, we explored the relationship between functional subregions and anatomical subregions by comparing anatomical position and connectivity patterns.ResultsBased on rsFC patterns, we parcellated the CC into six functional subregions :anterior cingulate cortex, ventral anterior mid-cingulate cortex, posterior midcingulate cortex, dorsal posterior cingulate cortex, ventral posterior cingulate cortex.Based on AC patterns, we parcellated the CC into ten anatomical subregions:subregion1(S1)-S10. Every subregion has its specific connectivity patterns. The functional subregions and the anatomical subregions had corresponding relationship.Anatomical S1 and S2 corresponded to anterior cingulate cortex, S3-6 corresponded to mid-cingulate cortex, S7 and S8 corresponded to dorsal posterior cingulate cortex,S9 and S10 corresponded to ventral posterior cingulate cortex.Conclusion1. The human CC can be divided into six separable functional subregions: anterior cingulate cortex, dorsal anterior mid-cingulate cortex, ventral anterior mid-cingulate cortex, posterior mid-cingulate cortex, dorsal posterior cingulate cortex, ventral posterior cingulate cortex.2. The human CC can be divided into ten separable anatomical subregions: S1-S10.3. There are corresponding relationship between the functional subregions and the anatomical subregions.