ISMRM 2015 Tractography challenge
The ISMRM 2015 Tractography Challenge seeked
to evaluate and compare state-of-the-art tractography pipelines.
The main bottleneck of tractography evaluation is the unavailability of groundtruths.
We overcome this limitation with a new data generation process
which allows to create a groundtruth for scoring using the
Tractometer online evaluation system.
The task of the challenge was to provide the best possible reconstruction of fiber pathways in a realistically simulated replication of a whole brain diffusion-weighted MR image with properties typical of a clinical acquisition.
The challenge is now closed. However, you can still download the data and work with it. In the coming weeks, a standalone evaluation tool will be released to allow scoring new submissions.
Data generation process in a nutshell
See here for a description of the data and the possibility to download it.
We have a mailing list to keep every contestant up to date on news, new results, etc.
To subscribe, click here or send an email to firstname.lastname@example.org with the following subject "SUBSCRIBE ismrm-2015-tracto-challenge" (without quotation marks).
Expected contestant work
The contestants were expected to process the challenge dataset using their best processing pipeline, as if they were asked to process the dMRI data of a clinical patient to produce the best possible tractography result. There were no constraints on the processing steps or running time, as we only expected to receive the tractogram file.
Peter Neher1, Jean-Christophe Houde2, Emmanuel Caruyer3, Alessandro Daducci2,4, Tim Dyrby5, Klaus Maier-Hein1, Bram Stieltjes6, Maxime Descoteaux2
- 1Medical Image Computing Group (MIC), German Cancer Research Center (DKFZ)
- 2Sherbrooke Connectivity Imaging Lab (SCIL), Université de Sherbrooke
- 3CNRS, IRISA, UMR 6074 VISAGES project-team, Rennes, France
- 4Signal Processing Lab (LTS5), Ecole Polytechnique Federale de Lausanne, Switzerland
- 5Diffusion Imaging Group, Danish Research Centre for Magnetic Resonance (DRCMR), Denmark
- 6University Hospital Basel, Radiology & Nuclear Medicine Clinic