RSNA Knee Abnormality Detection AI Challenge 2026
The RSNA Knee Abnormality Detection AI Challenge dataset pairs knee MRI examinations with their original multilingual radiology reports. It was created for developing models that detect and classify knee abnormalities using both imaging and report text.
Access is provided through the RSNA competition hosted on Kaggle and requires a Kaggle account and acceptance of the competition rules. Because competition data are governed by challenge-specific terms rather than a general open-data license, commercial reuse should be treated as restricted unless the current rules explicitly permit it.
RSNA reports more than 5,000 knee MRI examinations collected from 19 sites worldwide, with associated radiology reports written in 12 languages. An exact examination count, unique-patient count, individual image count and definitive downloadable file-format list were not published on the official overview page, so those fields have not been guessed. This is an active 2026 competition dataset and its access conditions may change after the challenge closes.
This record is a pointer. Licence and access terms are recorded as they stood on the verification date above; what binds you is the dataset's own licence text. Confirm at the source before use.