Neurofibromatosis Tumor Segmentation on Whole-body MRI (WBMRI-NF) Challenge


The WBMRI-NF 2023 Challenge tasks participants to develop automated segmentation models of NF tumors on WBMRI.

Participation policies

The models submitted and to be considered for this challenge must be fully automated. Publicly available data are allowed for training. Members affiliated with the organizing committee may participate in the Challenge but not eligible for awards and not listed in leaderboard. The performance of all submitted models will be evaluated in terms of the weighted detection and segmentation metrices. The results will be announced publicly on a leaderboard, which will be updated daily during the open validation round. Each team has maximal 10 submissions (one submission per week) in the evaluation phase. Only the last run of the submitted Docker container is officially counted to rank challenge results. Each team requires to submit a short technical paper in the length of 4 pages. Otherwise, the team is not eligible for the final ranking and awards. The top 3 teams will be received cash prizes of 1000 USD. Top 5 teams will be invited to contribute to a full-length challenge summary research article.


All participant teams must submit their results in two phases. Open validation phases: the participating teams will be requested to send the results of their algorithms on the open validation dataset as a single compressed zip file via the Challenge submission system. Closed testing phases: the participating teams will be requested to submit their algorithms in the form of a Docker container to the Docker repository of the Challenge, accompanied with a short paper highlighting the contributions of the submission. The Docker container will be required to follow particular guidelines in order to be handled by the challenge scoring infrastructure. After the Challenge, the Docker models will be released with the summarized paper of the Challenge.