Characterizing functional modules in the human thalamus: coactivation-based parcellation and systems-level functional decoding

Abstract

The human thalamus relays sensory signals to the cortex and facilitates brain-wide communication. The thalamus is also more directly involved in sensorimotor and various cognitive functions but a full characterization of its functional repertoire, particularly in regard to its internal anatomical structure, is still outstanding. As a putative hub in the human connectome, the thalamus might reveal its functional profile only in conjunction with interconnected brain areas. We therefore developed a novel systems-level Bayesian reverse inference decoding that complements the traditional neuroinformatics approach towards a network account of thalamic function. The systems-level decoding considers the functional repertoire (i.e., the terms associated with a brain region) of all regions showing co-activations with a predefined seed region in a brain-wide fashion. Here, we used task-constrained meta-analytic connectivity-based parcellation (MACM-CBP) to identify thalamic subregions as seed regions and applied the systems-level decoding to these subregions in conjunction with functionally connected cortical regions. Our results confirm thalamic structure–function relationships known from animal and clinical studies and revealed further associations with language, memory, and locomotion that have not been detailed in the cognitive neuroscience literature before. The systems-level decoding further uncovered large systems engaged in autobiographical memory and nociception. We propose this novel decoding approach as a useful tool to detect previously unknown structure–function relationships at the brain network level, and to build viable starting points for future studies.

Molecular signatures of attention networks

“…The analysis of subcortical areas similarly failed to establish a clear relationship between ANT activations and molecular measures. Nevertheless, it is essential to recognize that subcortical nuclei are often relatively small and may not be adequately distinguished at the limited spatial resolution inherent to transcriptomic, molecular, and functional neuroimaging techniques (for an in‐depth discussion, please refer to Boeken et al, 2022). Future investigations may consider optimizing processing pipelines for subcortical areas or employing spatially constrained hypotheses to uncover more regionally specific effects.…”

fMRI-based spatio-temporal parcellations of the human brain

“…The concept of clustering the time courses of voxels serves as a fundamental strategy in brain parcellation, aiming to aggregate brain units with similar features into clusters while ensuring distinct differences between these clusters. Several clustering techniques have been developed for this purpose, including k-means [20 ▪ ], hierarchical clustering [21], spectral clustering [22], and the region-growing method [23]. While these methods have been instrumental in creating both traditional atlases and novel brain parcels, they face the challenge of selecting the optimal number of clusters – a critical decision on “lumping vs. splitting”, that significantly influences the outcomes.…”

Multimodal imaging investigation of structural rich club alterations in Alzheimer's disease and mild cognitive impairment: Amyloid deposition, structural atrophy, and functional activation differences

“…Functional characterization of all 82 regions of interests was achieved via meta‐analytic reverse inference decoding, using the NeurosynthDecoder as implemented in the Neuroimaging Meta‐Analysis Research Environment (NiMARE; Salo et al, 2022), following our standard protocols (Boeken et al, 2022; Boeken & Markett, 2023). Neurosynth is a comprehensive data base of the functional neuroimaging literature and contains functional activation foci from 14,371 functional neuroimaging together with relevant meta data (called terms) which can be queried to gain insights under which conditions these activation foci have been obtained.…”