A data-driven approach to mapping cortical and subcortical intrinsic functional connectivity along the longitudinal hippocampal axis.

The hippocampus (HPC) is functionally heterogeneous along the longitudinal anterior-posterior axis. In rodent models, gene expression maps define at least three discrete longitudinal subregions, which also differ in function, and in anatomical connectivity with the rest of the brain. In humans, equivalent HPC subregions are less well defined, resulting in a lack of consensus in neuroimaging approaches that limits translational study. This study determined whether a data-driven analysis, namely independent component analysis (ICA), could reproducibly define human HPC subregions, and map their respective intrinsic functional connectivity (iFC) with the rest of the brain. Specifically, we performed ICA of resting-state fMRI activity spatially restricted within the HPC, to determine the configuration and reproducibility of functional HPC components. Using dual regression, we then performed multivariate analysis of iFC between resulting HPC components and the whole brain, including detailed connectivity with the hypothalamus, a functionally important connection not yet characterized in human. We found hippocampal ICA resulted in highly reproducible longitudinally discrete components, with greater functional heterogeneity in the anterior HPC, consistent with animal models. Anterior hippocampal components shared iFC with the amygdala, nucleus accumbens, medial prefrontal cortex, posterior cingulate cortex, midline thalamus, and periventricular hypothalamus, whereas posterior hippocampal components shared iFC with the anterior cingulate cortex, retrosplenial cortex, and mammillary bodies. We show that spatially masked hippocampal ICA with dual regression reproducibly identifies functional subregions in the human HPC, and maps their respective brain intrinsic connectivity. Hum Brain Mapp 37:462-476, 2016.