Instruction to participate¶

• Create an account on Grand Challenge, join our challenge and download the training data from the Dataset page.
• Submit your solution by docker container to the preliminary test phase during the opening period to validate your solution. Each participant is allowed up to 16 submissions within this period.
• Submit your final solution by docker container for the final test phase. Each participant is allowed only 1 submission. Please submit a 2-3 page summary detailing your methods as well. 
• In both preliminary test and final test phases, the solutions participants submit are supposed to be the segmentation models after finetuning, which can segment diffrent biological structures.  

💡Notes:¶

• Please refer to our SELMA3D_challenge-submission-example repo on GitHub as a template and guide for submissions to the preliminary test phase by docker container.
• Please refer to our SELMA3D_challenge-submission-example-for-final-test-phase repo on GitHub as a template and guide for submissions to the final test phase by docker container.
• Please remember to verify your account before submission.
• The preliminary test phase is not used for final evaluation. Please use preliminary test phase to debug and validate the docker submission of your solution.
• The two test phases take 1 or 2 hours to evaluate. Plan accordingly and avoid last-minute submissions for the final test phase.

Usage of Training Data¶

• Participants must use only the provided training data, which consists of two subsets. The first subset includes unannotated whole-brain 3D LSM images for self-supervised learning of models. The second subset contains annotated patches for model finetuning. Importantly, these annotations are on different structures from those in the validation and test sets. This is to make sure that participants develop self-supervised methods.
• We provide a validation set (without annotations) to illustrate the samples for the preliminary test phase. Participants must not manually annotate these images to train a supervised model for submissions during this phase.

Members of the organizers' institutes¶

• Participants from the organizers' groups are welcome to join and can have their results featured in publications and on the leaderboard. However, they are not eligible for awards.

Award and Result Announcement Policy¶

• All results will be disclosed publicly, and outstanding submissions will be acknowledged during the in-person event in MICCAI 2024.
• The top 5 participants will be invited to prepare a 10 minute presentation for the challenge session to showcase and discuss their methods.
• The top 3 participants will be publicly acknowledged on this challenge webpage, and they will receive a Jellycat Brezel 🥨 as a souvenir during the in-person challenge event.
• After the public acknowledgement, a detailed analysis of the submitted results will be available upon request.

Publication policy¶

We will collaborate with participants to create a comprehensive journal article summarizing the key results and analyses from this challenge. Participants who submit valuable work are welcome to contribute to the publication , with up to three authors from each team being acknowledged.

In order for us to include you in our paper:

• Please submit a detailed description of your solution with your final test phase submission.
• You are welcome to submit additional paragraphs and figures about your submission via email (see the Contact info on the Organizers page).

Besides, we encourage all participants to independently submit their results without imposing any publication embargo.