CAROT

CAROT maps atlas-parcellated neuroimaging time series between brain atlases using Optimal Transport to produce comparable connectomes without requiring access to raw data.

Key Features:

• Optimal Transport mapping: Uses Optimal Transport theory to establish mappings between distinct brain atlases.
• Atlas-to-atlas remapping without raw data: Transforms data processed with one atlas into a connectome based on another atlas without requiring access to raw voxelwise data.
• Connectome reconstruction: Reconstructs connectomes that closely resemble connectomes generated directly from target atlases.
• Downstream analysis compatibility: Produces transformed connectomes suitable for downstream analyses, including connectome-based predictive modeling.
• Predictive model generalization: Facilitates generalization of connectome-based predictive models to datasets processed with different atlases.
• Reduces redundant preprocessing: Enables reuse of preprocessed atlas-parcellated datasets across atlas frameworks.

Scientific Applications:

• Cross-atlas connectome harmonization: Enables comparison and integration of connectomes derived from different atlases for connectomics studies.
• Connectome reconstruction and validation: Provides reconstructed connectomes for assessing fidelity to atlas-derived originals.
• Transferability of predictive models: Permits applying connectome-based predictive models across public neuroimaging datasets processed with different atlases.
• Maximizing reuse of preprocessed data: Increases the utility of atlas-parcellated neuroimaging datasets for diverse analytical workflows.

Methodology:

Establishes mappings between atlases using Optimal Transport and applies those mappings to atlas-parcellated voxelwise time series to reconstruct target-atlas connectomes without access to raw data.