RSNA Knee MRI AI Challenge (2026)
The 2026 RSNA Knee MRI AI Challenge will engage researchers to develop machine learning models to detect abnormalities.
The challenge will be the first to use both images and the text of radiology reports to train and test AI models. The competition will launch in Spring 2026 and conclude in October.
Why Is MRI Essential to Evaluating Knee Injuries?
MRI is central to evaluating knee injuries because it reveals damage to joint structures, including ligaments, menisci and cartilage loss, providing a comprehensive whole-joint assessment. MRI can also detect abnormalities such as bone marrow lesions, effusion, synovitis and Baker cysts that cause symptoms and require management.
However, MRI interpretation varies across readers and practice settings and access to specialty-trained musculoskeletal radiologists is limited. By training AI models on a large multilingual dataset of knee MRIs paired with their original reports, this challenge aims to develop tools that can reliably identify abnormalities and support more consistent, timely care.
Volunteer to Annotate and Review
RSNA invites trained musculoskeletal radiologists to participate in the challenge by annotating knee MRI studies and reviewing associated reports. The work requires approximately 10 hours of commitment between February and March 2026. Those who complete the assignment will be recognized in associated research publications and acknowledged in AI Challenge communications.
Your expertise helps create a one-of-a-kind dataset with over 5,000 knee MRI exams from 16 institutions worldwide, paired with corresponding MRI reports in nine languages!
Contact Us
For more information, contact us at informatics@rsna.org.