🤝 Obligations of Participants
To qualify for ranking, awards, or any honors in this Challenge, participants must meet the following obligations:
- Present their research methods at the final event of the Challenge held during the 2026 International Symposium on Biomedical Imaging (ISBI).
- Participants must sign and comply with the Data Use Agreement.
- Submit a paper that reports the details of the research methods, with a length limit of no more than 5 pages and include a link to the code used for training.
- Sign and return all prize confirmation documents required by the competition sponsors/organizers.
- Participants commit to citing the papers related to this Challenge and the data overview paper when submitting their developed methods to scientific or non-scientific publications, and any publication using the Challenge data must wait until the Challenge overview paper has been published.
- Ensure that the dataset provided in this Challenge is not used for commercial purposes.
📒 Accounts
All participants must register for the Challenge using a single, verified account under their real name. Registration must include full affiliation details (department, institution or company name, and country) and an official institutional email address. Anonymous participation and the use of multiple accounts by any individual are strictly prohibited. Only accounts that satisfy these criteria are eligible to make submissions. The organizing committee reserves the right to disqualify any participant who violates these provisions from inclusion on the validation or testing leaderboards.
🏟 Teams
Teams are allowed to participate, but team members are not permitted to submit individually outside their team, as doing so would unfairly increase the number of submissions. Each participant may join only one team. Individuals found using multiple or duplicate accounts will be disqualified. Participants from the same research group are also prohibited from registering multiple teams. The organizers reserve the right to disqualify any participant who violates these rules.
🏷️ Code-sharing
Code sharing is restricted to members within the same registered team. Any code exchanged with participants outside of one’s team must be made publicly available via the official Challenge forum. By submitting an entry, participants agree that their accompanying short papers may be published on the Challenge website. The organizers reserve the right to use materials submitted by participants, such as scores, predicted labels, and papers, for purposes related to the administration, promotion, and dissemination of the Challenge.
⚙️ Use of Training Data and Models
Participants are permitted to use publicly released pre-trained models. To maintain fairness and reproducibility, the use of non-public or unreleased pre-trained models is prohibited. Additionally, participants may not utilize any datasets other than those officially provided by the Challenge organizers.
👨💻 Eligibility
Individuals affiliated with the organizing institutions, including those involved in the design or acquisition of the training and test datasets, may participate in the Challenge. However, to prevent conflicts of interest, they are ineligible for awards and will not be ranked in the final testing phase.
🙅 Prohibit Ensemble Learning
Ensemble learning can improve performance by combining multiple base models, but it may mask flaws in model design and shortcomings in hyperparameter tuning. To more accurately evaluate participants’ abilities in core aspects such as feature engineering, model architecture design, and hyperparameter tuning, and to encourage focus on the performance and innovation of individual models, this rule is established: Participants employing ensemble techniques will be excluded from the leaderboard rankings.