A Precision Functional Atlas of Personalized Network Topography and Probabilities
This paper was published in the May 2024 issue of Nature Neuroscience, with authors from institutions including the University of Minnesota. The study aimed to address individual differences in brain functional networks by establishing an open-source “Mapping Individual Differences in Brain Networks” (MIDB) resource.
Research Background:
Although brain functional networks share overall commonalities, their spatial topological structures exhibit substantial inter-individual differences. Traditional group-averaged network maps ignore these individual differences, potentially reducing statistical power in large-scale studies and precision in neuromodulation therapies. Therefore, brain map resources that precisely describe individual network topologies are needed.
Data Sources: The authors utilized resting-state functional MRI data from the Adolescent Brain Cognitive Development (ABCD) study, which recruited nearly 12,000 9-10-year-old adolescents and plans to follow them for 10 years. Additionally, data from the Human Connectome Project Development (HCP-D) and other datasets were used.
Research Procedure: 1) Multiple network detection algorithms (Infomap, template matching, non-negative matrix factorization, etc.) were employed to map individual-specific networks for each participant. 2) At the group level, the probability of each gray matter voxel being assigned to different networks was calculated to generate probabilistic network maps. 3) An “Overlapping Multi-Network Imaging” (OMNI) method was proposed, allowing network overlap to reveal integrative regions. 4) The role of these precise network maps in improving replicability and predictive power in large-scale studies was evaluated.
Key Findings: 1) Individual network mapping results showed good reproducibility across different algorithms and data sizes. 2) Group-level probabilistic network maps exhibited high consistency across age groups, algorithms, and when including task data. 3) High-confidence regions extracted from probabilistic network maps improved reproducibility and predictive power in brain-wide association analyses. 4) OMNI mapping revealed “integrative regions” in the brain, potentially playing a key role in information integration. 5) Probabilistic network maps facilitated more precise target selection for neuromodulation therapies.
Significance: 1) Scientific: Elucidating individual and group-level differences in brain functional network topologies. 2) Application: Providing resources for individualized neuroimaging and neuromodulation therapies, enhancing analysis and treatment precision.
Research Highlights: 1) Large sample size, spanning adolescents to adults. 2) Advanced individual network mapping and group probabilistic clustering algorithms. 3) Open-source data resources, contributing to the scientific community.
This study systematically establishes a precise map of individual and group-level brain functional network topologies, demonstrating its potential applications in improving replicability in large-scale studies and targeted neuromodulation. It represents an important resource for functional neuroimaging and individualized neuroscience research.