Mapping higher-order language regions in the cerebellum with precision functional MRI

1016 

Abstract Submission 

Joseph James Salvo¹, Nathan Anderson¹, Rodrigo Braga¹

¹Northwestern University, Chicago, IL

Joseph James Salvo  
Northwestern University
Chicago, IL

Nathan Anderson  
Northwestern University
Chicago, IL

Rodrigo Braga  
Northwestern University
Chicago, IL

The brain is populated by large-scale networks that interconnect regions in the cerebrum and cerebellum (Buckner et al. 2011; Xue et al. 2021). Supporting specialization at the level of the entire cerebral-cerebellar networks, estimates of language task-related activity reveal that perisylvian cerebral areas activate alongside cerebellar regions during linguistic tasks (Petersen et al., 1989; Ashida et a. 2019). Recently, within-individual estimates from resting-state functional connectivity have outlined a set of cerebellar regions that putatively belong to a distributed language network (Xue et al., 2021; King et al., 2019). Here, we tested the correspondence between these cerebellar language network regions and regions showing task-evoked responses during auditory and visual language tasks.

Extensive task and resting-state 3T fMRI data were collected from 8 neurotypical adults over the course of 8 sessions, using a multi-echo protocol. Participants were presented with written (total 40 min; Fedorenko et al., 2010) or spoken (total 48 min; Scott et al., 2017) sentences, as well as control (contrast) conditions including lists of written pronounceable pseudowords or distorted speech, respectively. Participants also provided 16 resting-state runs (total 112 min).

Runs containing head motion were excluded if max framewise displacement (FD) > 0.4mm, and/or max absolute displacement > 2mm, or were visually checked for exclusion if max FD > 0.2 and/or max absolute displacement > 1mm. We verified that each run's field of view contained the full cerebellum. Included runs were aligned to a T1 template and projected to a standardized cerebral (Fischl et al., 1999) and cerebellar surface (Diedrichsen & Zotow, 2015). Functional connectivity was used to estimate the language network in the cerebrum based on Braga et al. (2020), using manually chosen seeds and data-driven clustering. Cerebellar vertices were assigned to networks defined on the cerebrum using a winner-take-all approach based on functional connectivity (Xue et al., 2022).

Functional connectivity reliably defined multiple language network regions in the cerebellum. In some participants, these regions were right-lateralized (i.e., larger on the right hemisphere), while in others we observed bilateral regions. Across participants, we consistently observed 3-4 distinct regions in the posterior cerebellum, rather than one contiguous region. Despite individual differences, we in general observed correspondence between the functional connectivity estimates and task-related activity for both the auditory and visual language tasks. Finally, in between the distinct regions we observed other networks, suggesting fine-scale interdigitation between networks when individual-level anatomy is considered. Finally, mirroring our observations in the cortex, we observed that the auditory task led to more bilateral cerebellar activity in most participants, which could have been driven by recruitment of auditory processing regions.

The results suggest that resting-state functional connectivity-based estimates of the language network can delineate language-task-responsive regions of the cerebellum. Further, the results suggest that these language network regions respond similarly to spoken or written language. Ongoing work is exploring the relationship of these higher-order language regions with sensory input streams.

Language Comprehension and Semantics ¹

Reading and Writing

Activation (eg. BOLD task-fMRI) ²

fMRI Connectivity and Network Modeling

Cerebellum

Cortex

FUNCTIONAL MRI

Language