Cross Atlas Remapping via Optimal Transport (CAROT): Creating connectomes for different atlases when raw data is not...

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Data Science

Background:

• Open-source neuroimaging datasets promote scientific reproducibility and discovery by offering large sample sizes.
• Current practices often release preprocessed data parcellated into predefined brain atlases, limiting downstream analysis flexibility.
• Connectomes derived from different atlases are incomparable, hindering the integration of diverse open-source datasets.

Purpose of the Study:

• To develop a method for transforming neuroimaging data between different brain atlases.
• To enable the direct comparison and integration of connectomes processed using disparate atlases.
• To overcome limitations in the reusability of open-source neuroimaging data.

Main Methods:

• Introduced Cross Atlas Remapping via Optimal Transport (CAROT) to establish mappings between brain atlases.
• Developed a computational approach to transform preprocessed connectomes from one atlas space to another.
• Validated the method by comparing reconstructed connectomes and assessing downstream analysis performance.

Main Results:

• CAROT successfully reconstructs connectomes across a wide range of atlases without requiring raw data.
• Transformed connectomes demonstrated utility in downstream analyses, including predictive modeling.
• Connectome-based predictive models generalized to data processed with different atlases after transformation.

Conclusions:

• CAROT facilitates the seamless integration and reuse of existing processed neuroimaging connectomes.
• The tool democratizes access to connectomic data by overcoming atlas-related comparability issues.
• CAROT promotes broader scientific collaboration and discovery by enhancing the utility of open-source brain data.