MICCAI 2022 Daily – Tuesday

24 DAILY MICCAI Tuesday Challenge - FeTS by Spyridon Bakas, Maximilian Zenk, Ujjwal Baid Spyridon is Assistant Professor at the University of Pennsylvania and Board Member of the MICCAI Special Interest Group on Biomedical Image Analysis ChallengeS (SIG-BIAS) Towards the next generation of biomedical image analysis challenges! The FeTS challenge represents the first federated machine learning challenge ever proposed across domains, and the first of what-we- hope-to-be the next generation of computational competitions (i.e., challenges) in healthcare, following a clinical trial design. Its first instance was conducted in conjunction with MICCAI 2021 and, following its success and community engagement, it was proposed and invited to run again during MICCAI 2022. Taking into consideration the readership of the Computer Vision News magazine, I would like to clarify what is meant by “ clinical trial design ” in the context of challenges. Following a clinical task definition, biomedical image analysis challenges so far have only been focusing on evaluating feasibility (i.e., phase 1 clinical trials), where the performance of developed algorithms is quantitatively assessed on a small test set under ideal conditions, e.g., well-curated centralized data. With the FeTS challenge we decided to go beyond this phase 1 (feasibility evaluation) and quantitatively evaluate the capability of algorithms in an environment simulating real-world environments (i.e., phase 2 clinical trials) by accessing multi-institutional shared data, and importantly evaluate the effectiveness of algorithms in real-world clinical environments (i.e., phase 3 clinical trials) that did not contribute