A connectivity-based parcellation improved functional representation of the human cerebellum

Abstract

The cerebellum is traditionally well known for its role in motor learning and coordination. Recently, it is recognized that the function of the cerebellum is highly diverse and extends to non-motor domains, such as working memory, emotion and language. The diversity of the cerebellum can be appreciated by examining its extensive connectivity to the cerebral regions selective for both motor and cognitive functions. Importantly, the pattern of cerebro-cerebellar connectivity is specific and distinct to different cerebellar subregions. Therefore, to understand the cerebellum and the various functions it involves, it is essential to identify and differentiate its subdivisions. However, most studies are still referring the cerebellum as one brain structure or by its gross anatomical subdivisions, which does not necessarily reflect the functional mapping of the cerebellum. We here employed a data-driven method to generate a functional connectivity-based parcellation of the cerebellum. Our results demonstrated that functional connectivity-based atlas is superior to existing atlases in regards to cluster homogeneity, accuracy of functional connectivity representation and individual identification. Furthermore, our functional atlas improves statistical results of task fMRI analyses, as compared to the standard voxel-based approach and existing atlases. Our detailed functional parcellation provides a valuable tool for elucidating the functional diversity and connectivity of the cerebellum as well as its network relationships with the whole brain.

Intrinsic Functional Connectivity is Organized as Three Interdependent Gradients

“…S1 and MNI coordinates in Table S1. Homogeneity of each seed was calculated as the average temporal correlation between all unique pairs of within-seed voxels following previous reports 18,81,150,151 . The mean (M) and standard deviation (SD) of seed homogeneity for each network motif was identified as follows: Attention (M = 0.800, SD = 0.019), default (M = 0.822, SD = 0.016), executive (M = 0.810, SD = 0.016), exteroceptive (M = 0.795, SD = 0.023), and salience (M = 0.797, SD = 0.016).…”

Cerebellar network organization across the human menstrual cycle

“…Further, we predicted estradiol would augment cerebellar coherence. For each session, the cerebellum was parcellated into 99 nodes and then were spatially mapped to a seven-network atlas 13,45 . A summary time-course was extracted from each node, data were temporally filtered, and 99 × 99 functional association matrices were derived via magnitude-squared coherence (FDR-thresholded at q < 0.05; see "Methods and materials" for full description of preprocessing and connectivity estimation).…”

Dynamic cerebellar network organization across the human menstrual cycle

“…Further, we predicted estradiol would augment cerebellar coherence (Pritschet et al, 2019). For each session, the cerebellum was parcellated into 99 nodes from the Ren atlas that were then mapped to the Buckner seven-network atlas (Ren et al, 2019;Buckner et al, 2011). A summary time-course was extracted from each node, data were temporally filtered, and 99 x 99 functional association matrices were derived via magnitudesquared coherence (FDR-thresholded at q < .05; see Methods and Materials for full description of preprocessing and connectivity estimation).…”

“…All nuisance regressors were detrended to match the BOLD time series. Additionally, a 100-node atlas was preferred over other node atlas options (10 and 300 node) because it exhibited more moderate centering of each node and symmetry between the two hemispheres (Ren et al, 2019).…”