Date of Award
Fall 2023
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Computer Science
First Advisor
Scheinost, Dustin
Abstract
Open-source, publicly available neuroimaging datasets---whether from large-scale data collection efforts or pooled from multiple smaller studies---offer unprecedented sample sizes and promote generalization efforts. Releasing data can democratize science, increase the replicability of findings, and lead to discoveries. Due to patient privacy, computational, and data storage concerns, researchers typically release preprocessed data with the voxelwise time series parcellated into a map of predefined regions, known as an atlas. However, releasing preprocessed data also limits the choices available to the end-user. This is especially true for connectomics, as connectomes created from different atlases are not directly comparable. With the absence of gold standards for the numerous existing brain atlases, it is unrealistic to have processed, open-source data available for each. Together, these limitations directly inhibit the potential benefits of open-source neuroimaging data. In this thesis, I address these limitations and introduce Cross Atlas Remapping via Optimal Transport (CAROT) to find a mapping between two atlases that allows data processed from one atlas to be directly transformed into a connectome based on another atlas without the need for raw data access. To validate CAROT, I compare reconstructed connectomes against their original counterparts, demonstrate the utility of transformed connectomes in downstream analyses, and show how a connectome-based predictive model can generalize to publicly available data that were processed with different atlases. We show that CAROT can reconstruct connectomes from an extensive set of atlases ---without needing the raw data---allowing already processed connectomes to be easily reused in a wide range of analyses while eliminating redundant processing efforts. We also propose a benchmark based on the information cycle to investigate the amount of information lost in the reconstruction process. To this aim, several hypotheses were proposed as reference points, including the expectation of the lowest information loss when traversing within the same atlas, but with different resolutions, and extending to topologically similar and all atlases. CAROT exhibits varying levels of information loss depending on the target atlas highlighting the intrinsic features of the atlases as the main factor for variability. We then introduce a middle-way optimal transport to incorporate a third atlas when no mapping exists between the source and target. The results demonstrate that using existing mappings with a path between the source and target can achieve similar performance to direct transportation. The CAROT tool is shared as source code and a stand-alone web application at \url{http://carotproject.com/}.
Recommended Citation
Dadashkarimi, Javid, "Data-Driven Mapping Between Functional Connectomes Using Optimal Transport" (2023). Yale Graduate School of Arts and Sciences Dissertations. 1154.
https://elischolar.library.yale.edu/gsas_dissertations/1154
