Cohort-level brain mapping: learning cognitive atoms to single out specialized regions

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Mapping

Background:

• Functional Magnetic Resonance Imaging (fMRI) is crucial for mapping brain function by analyzing group responses to tasks.
• Current fMRI methods have limitations in the number of functional networks that can be extracted, which does not scale with cohort size.

Purpose of the Study:

• To introduce a new group-level brain mapping strategy to identify a greater variety of brain network configurations within a population.
• To overcome the limitations of existing fMRI analysis in delineating specialized brain regions and networks.

Main Methods:

• A novel group-level strategy based on functional segregation is proposed.
• Dictionary learning with on-line algorithms is employed for efficient, sparse representation of brain region responses.
• Inter-subject correspondence is modeled using structure-inducing regularization within functional profiles.

Main Results:

• The approach successfully extracts a large number of brain networks from large multi-subject fMRI data.
• The identified functional profiles are meaningful and reflect diverse brain network configurations.
• The method scales efficiently to large datasets without additional computational cost for regularization.

Conclusions:

• The developed group-level brain mapping strategy enhances the ability to differentiate numerous brain regions and networks.
• This method offers a scalable and computationally efficient approach for analyzing complex brain network configurations in large cohorts.
• The findings advance the understanding of population-level brain organization using fMRI data.