AI That Reads Knee MRIs Could Speed Up Osteoarthritis Treatments

Researchers have developed an AI tool that automatically measures knee cartilage from MRI scans, the standard way doctors track osteoarthritis. Osteoarthritis is the most common form of arthritis, causing knee pain, stiffness, and swelling as the cartilage that cushions joints wears away. Because cartilage changes are tiny and slow, measuring them by hand is tedious and inconsistent. This AI does the measuring automatically, like a super-precise digital ruler.

The system was built and tested using data from a large, multi-center (multiple hospitals) Phase III clinical trial—the final stage of drug testing in humans. The AI first outlined the knee cartilage on its own, and then two expert readers reviewed and corrected its work. A third reader stepped in if they disagreed. The whole process was applied to almost 1,200 real MRI exams from over 400 patients.

The results were impressive. The AI's outlines matched the expert-corrected outlines about 96% of the time, and different readers agreed with each other almost perfectly. The AI also passed a "synthetic thinning" test, where researchers artificially shrank the cartilage on 20 computer-generated knee models to see if the AI could detect the change. It did so with high accuracy.

But the real test came when the researchers compared two groups in the trial: patients receiving an experimental treatment and patients receiving a placebo. Over an 8-month period, the AI detected that the treated group's knee cartilage increased in volume and thickness while the untreated group's decreased. That means the AI is sensitive enough to catch real biological changes—making it a valuable tool for proving whether new arthritis drugs actually work. This could lead to faster, more reliable clinical trials and, ultimately, better treatments for people with knee osteoarthritis.