Tractography - Randomized filtering and Redundancy quantification

Tractography in diffusion MRI is widely used for mapping brain connectivity but often suffers from implausible and redundant streamlines, leading to possible inaccuracies in downstream analyses. This seminar presents my first year of work on addressing these challenges through two approaches and future research directions. First, Randomized COMMIT refines tractograms by iteratively optimizing streamline selection on tractogram subsets. Second, we have introduced a statistical framework that quantifies tractogram redundancy, providing insights into streamline plausibility and representation.

Reviewers: Zhikai Yang and Eva Breznik