SEG.A. - Segmentation of the Aorta Banner
Join

Questions?

  • I can not use the docker and find the test data.

    Unlike previous years, the challenge organization is to be found in "algorithm submission." This means that participants need to submit a docker container of their algorithm on the challenge website, and the algorithm will be executed remotely on a hidden test set. We provided a template for such a docker container, where the participants only need to modify only the segmentation method in the SegaAlgorithm/process.py file. As such, the test set will remain hidden for the rest of the challenge. Instructions on how to work with algorithm submissions can be found in the grand challenge documentation here.

    This first phase of the challenge is for the participants to get familiar with the submission system and be ready for the final phase. Suppose you can't run docker on your local machine. In that case, you can still edit the segmentation generation part in the template container and submit it (up to 5 times) to test it.

  • I have my Docker image, but I do not know the hardware specification of the docker.

    According to the Grand Challenge documentation, the nodes use a GPU NVIDIA T4 with 16GB of memory. At the end of the documentation page of the grand-challenge website (this page), there's a complete description of this topic.

  • Where can I find the test cases?

    The test cases are embedded in the docker container of the challenge. Therefore, the cases will remain hidden for the time being.

  • Yes, top-performing teams will have the opportunity to be listed as authors in the challenge paper. To participate, you must submit a 10-12 page article through CMT3, following the LNCS style guidelines similar to MICCAI workshops. Additionally, we will write a challenge summary paper with the support of participants, who will be invited as co-authors. We aim to involve as many teams as possible, typically with 2-3 authors per team, depending on the number of submissions and participants. We are still collecting information about articles possibilities and authorship. Further information will follow.