The Martinos Center for Biomedical Imaging, the University of Rochester, and medical image processing company QMENTA have concluded the groundbreaking IronTract Challenge, which brought together bioimaging researchers across the world to collaborate on and establish an objective assessment of accuracy of computational algorithms for Diffusion MRI Tractography compared to ground truth histological data.
Diffusion MRI Tractography is a technique for reconstructing the white-matter bundles that form the wiring of the brain. This technique, which is used in assessment of neurological and psychiatric conditions as well as examining brain development, is currently generating hundreds of thousands of high-quality datasets across many active studies worldwide. Tractography is the only noninvasive and in vivo method for mapping the wiring of the brain, but is hampered by concerns related to measurement accuracy. Emerging AI algorithms have significant potential to improve tractography accuracy, but the ability to measure the accuracy of the algorithms themselves is key for validation and improvement.
Under the partnership of the Martinos Center for Biomedical Imaging, the University of Rochester and QMENTA, a community of groundbreaking computational researchers working on diffusion MRI came together under the IronTract Challenge. The challenge used a unique dataset not previously available to the research community, which consisted of histological ground truth on brain connections from Dr. Suzanne Haber’s laboratory at the University of Rochester Medical Center, and high-resolution diffusion MRI data acquired at the Martinos Center for Biomedical Imaging under the supervision of Dr. Anastasia Yendiki. These data were made available on QMENTA’s cloud platform. The global research community successfully used the platform to share/port their cutting-edge tractography algorithms and preprocessing workflows, measure their accuracy versus ground truth, and collaborate to advance the knowledge of the research community on the great challenges related to accuracy, reproducibility and standardization of diffusion MRI reconstruction methods.
QMENTA is awarding prizes to the researchers who achieved high algorithmic accuracy, and ported their algorithms to the QMENTA platform, to be shared for the benefit of the broader research community, especially applied neuroscience groups that don’t have the support of large IT departments or technicians to run scripts and debug code. With QMENTA these complex workflows can be run by simply selecting the datasets and clicking play to start the analysis.
The Martinos Center for Biomedical Imaging, the University of Rochester, and QMENTA are delighted to announce the prize recipients:
1st prize: G. Girard, Radiology Department, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland, Signal Processing Lab (LTS5), École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
2nd prize: B. Aydogan, Department of Neuroscience and Biomedical Engineering, Aalto University, Helsinki, Finland
3rd prize: M. Mancini, Department of Neuroscience, Brighton and Sussex Medical School, University of Sussex, Brighton, United Kingdom, NeuroPoly Lab, Polytechnique Montreal, Montreal, QC, Canada
In early 2021 the Martinos Center for Biomedical Imaging, the University of Rochester, and QMENTA plan to jointly host a webinar to formally celebrate the winners, share more insights from the challenge, explore how developers can make their tools seamlessly available to researchers, and how researchers can further expand their collaboration and improve the execution of their research studies.