Grant: Discovery Grants Program - Individual

TractoMap: A Novel Computational Framework for Analyzing White Matter Pathways with Diffusion MRI Tractography is a technique that can estimate white matter (WM) pathways in the brain from a non-invasive diffusion Magnetic Resonance Image (dMRI). However, computational challenges persist in characterizing WM connections due to the brain's complexity, the large number of generated pathways, and the high-dimensional nature of tractography reconstruction. Existing approaches, predominantly based on connectivity matrices, often oversimplify and do not sufficiently capture the complexity of WM pathways. This research aims to develop a computational framework, algorithms and data structures to better organize WM pathways reconstructed by dMRI tractography. The objective is to evaluate, enhance and build distance metrics, spatial partitioning techniques, and search methods designed specifically for tractography `streamlines', which represent WM pathways. These methods will improve how we search, compare, and categorize brain connections, aiming to establish a comprehensive tree structure. The resulting Tractography Map (TractoMap) combined with efficient search algorithms will allow us to study intricate structural patterns of streamlines, ranging from small to large fascicles (i.e. isolated to densely connected nodes or subgraphs in a network). To address this, spatial indexing techniques and tractography-specific search methods will be evaluated. Various distance measures will be designed and compared to existing ones, enhancing the identification and characterization of WM trajectories. The second focus is on creating a hierarchical representation of WM structures, involving the formulation of a space partitioning tree that efficiently groups similar trajectories. Utilizing dMRI data from large databases, such as Human Connectome Project and UK BioBank, we aim to build TractoMap, offering a data-driven template of human brain connectivity. This map will be aligned with a common space, such as the Montreal Neurological Institute (MNI) brain atlas, in order to promote its widespread use. The final emphasis will be on reproducibility and variability assessment, involving validation procedures to guarantee the reliability of tractography algorithms and brain connectivity analyses. With TractoMap as a fundamental reference, WM structures will be rigorously compared across multiple dMRI tractography reconstructions, incorporating various variables like demographics and pathology to enrich our understanding of the human brain connectivity. Resulting computational tools with the proposed trajectory-based representation will improve structural connectivity analysis. For equitable access and widespread impact, all data structures, algorithms, and the TractoMap atlas created during this research will be open-sources. To ensure reproducibility and facilitate collaborative research, we also plan to integrate these algorithms into established open libraries.