Tractography reproducibility challenge with empirical data (TraCED): The 2017 ISMRM diffusion study group challenge.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Computational Neuroscience

Background:

• Diffusion-weighted MRI (DW-MRI) is crucial for mapping in vivo brain white matter architecture.
• Tractography results are highly sensitive to processing methods and tracking parameters.

Purpose of the Study:

• To evaluate the reproducibility and variability of tractography algorithms in group studies.
• To address the limitations of phantom studies by incorporating in vivo physiological factors.
• To establish a benchmark for algorithm performance through the ISMRM 2017 TraCED challenge.

Main Methods:

• A systematic review and comparative analysis of tractography algorithms.
• Utilized data from single healthy volunteers acquired on two 3.0T scanners.
• Assessed 46 tractography algorithm entries using intraclass correlation (ICC) and Dice similarity coefficient.

Main Results:

• The top five algorithms demonstrated high reproducibility (ICC, Dice >0.88) across sessions and scanners (0.87-0.97).
• Containment analysis confirmed that top-performing algorithms' results were consistent with larger volume tracts.
• Out of 16 evaluated tracts, 8 showed high reproducibility, 4 moderate, and 4 low.

Conclusions:

• Tractography algorithms yield fundamentally different white matter structures, particularly with conservative specificity settings.
• The ISMRM 2017 TraCED challenge data and infrastructure provide a valuable platform for ongoing algorithm comparison and development.