This is intended as an algorithmic challenge with particular focus on techniques which leverage the combined information of both PSMA and FDG scans at the patient level. As such we request that participants do not utilise any external training data in their techniques. The organisers have provided CT-based organ delineations from Total Segmentator and these will be available at time of validation & testing evaluation but no other pre-trained models should be incorporated.

The top five performing entries and best multi-modality entry (if different from top 5) will be invited to participate in the DEEP-PSMA MICCAI workshop session and be included in the challenge manuscript.

The other rules for this challenge should be familiar to those who have participated in other Grand Challenge events:

• Best Multi-modality Algorithm Award: it is anticipated that many submissions will evaluate the PSMA and FDG PET/CT scans in isolation, however, a separate award is reserved for the best-performing entry which makes meaningful use of the combined PSMA and FDG dataset at the patient level; that is, it utilises some information in the individual's PSMA scan to improve performance in the FDG inference and vice versa. If your algorithm should be considered for this award, please include a note in your algorithm description. If you're uncertain about whether a potential method might be eligible prior to development and would like clarification please reach out to the organiser team and we will advise in confidence.

• Members of the organising institution as well as collaborators with challenge organisers are not eligible for final leaderboard or awards.

• Testing submission must be made by automated Docker container to be run in an offline environment.

• By participating, users agree to make their models available for future evaluation related to the challenge including methodological summary paper.

• In order to be eligible for the final test data leaderboard, participants are asked to provide a 1-2 page description of their algorithm methodology and a link to the training source code, preferably via GitHub repository.

• Access terms for ongoing use of the DEEP-PSMA challenge training dataset by attribution are being finalised and is planned to become open access shortly after challenge completion.

• All participants reserve the right to withdraw from the challenge and forego further participation. However, they will not be able to retract their prior submissions or published results until then.

• Participants are invited to prepare their own methodological publications based on results and algorithms developed in the DEEP-PSMA challenge. We request that authors cite the primary challenge manuscript in any future publication.