RSNA launches new knee MRI AI challenge
The Radiological Society of North America on Wednesday launched its latest artificial intelligence challenge, seeking proposals on how to improve MR imaging of the knee.
RSNA is hoping to spur the creation of new tools to improve patient care, asking researchers to develop machine learning models that can detect and classify knee abnormalities.
The society said this is its first AI challenge that will use medical images, along with radiology reports, to train models. Previous contests have focused on intracranial aneurysms (2025), classifying lumbar spine degeneration (2024), and abdominal trauma detection (2023).
“What makes this year’s Knee MRI AI Challenge unique is that it is the most real-world challenge yet,” Po-Hao “Howard” Chen, MD, MBA, challenge co-leader and vice chair for AI at the Cleveland Clinic Diagnostics Institute, said in a statement Aug. 5. “Participants must learn from real-world diagnostic radiology reports, where findings are complex and answers are not neatly organized in a table. This brings the challenge closer to how clinical AI must actually be developed: by confronting the nuance and variability of real radiologic interpretation.”
RSNA notes that MRI is critical to evaluating knee injuries. However, interpretation can vary widely across practice settings, with potentially limited access to specialty-trained musculoskeletal radiologists. The society hopes that training AI models on these large data sets can improve access and support more consistent care.
This year’s training data includes over 5,000 knew MRIs and the related radiology reports from 16 healthcare sites across the globe. Expert radiologists have already annotated the data set used to assess model performance, RSNA noted. The challenge will assess competing researchers’ models on how well they perform at tasks including detection, localization and categorization of abnormal features.
All are welcome to take part in the challenge, with the competition conducted on a platform provided by Kaggle Inc. Top performing teams will share a total of $77,000 in prizes, including a new award for the most efficient models. Previous contests have drawn participation from thousands of experts.
This year’s challenge is running through Oct. 22, with winners slated to be announced at RSNA 2026, running Nov. 29 to Dec. 3 in Chicago.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.