The diffusion study group has had great success in advancing the field of fiber tracking through challenges based on numerical and physical phantoms. These studies provide concrete ground truths with which participants in the challenge (and beyond) can fairly compare their algorithms. Comparative analysis of diffusion fiber tractography with these datasets has become a cornerstone of our literature. Notably absent from this field is a large-scale reproducibility dataset that can be used to assess algorithms in the presence of empirical considerations. While there is general consensus in the research community that advanced models contribute meaningful biological information, specifically which technique is optimal or advantageous given practical considerations remains an active area of concern. Capturing empirical imaging DW-MRI considerations is challenging with simulations or phantom experiments given the propensity for patient motion, artifacts associated with fast imaging techniques, hardware changes/failures, and non-linear estimation processes. Innovations in understanding uncertainty of fiber tracking with advanced DW-MRI methods are essential for interpreting the growing, diverse body of literature on structural connectivity and its implications in neuroscience and medicine. The goal of the proposed TraCED challenge is to assess the reproducibility of common and emerging tracking pipelines/algorithms using clinically feasible imaging sequences.
This challenge will provide a snapshot of the current progress in the field through direct comparisons of tractography methods, preprocessing pipelines, artifact identification, robust processing, fiber orientation reconstruction, and consideration of imaging/physical space representation in the context of inter-scanner comparison. Notably, this challenge will not provide an assessment of absolute accuracy of fiber tracking methods, as the true brain fibers are unknown. Rather, this work is intended to complement simulation efforts and pre-clinical validation studies, which suffer from limitations in capturing physiological and imaging considerations of in vivo human studies. Extended virtual discussions will provide researchers with an opportunity to characterize their methods on a newly created and released standardized dataset of neuroanatomy on clinically acquired 3T MRI scanner. The datasets will be freely available both during and after the challenge. A group authored publication on the results of the challenge is planned for Magnetic Resonance in Medicine.