AI spots arthritis in routine bone-density scans

Published:

6 January 2025

40,000 UK Biobank participants’ bone-density images were used to create an automated system that could catch osteoarthritis before symptoms set in. 

An artificial intelligence tool, developed with bone-density images from around 40,000 UK Biobank participants, can spot osteoarthritis in scans that are routinely done in clinics to find fragile bones in older people.

Similar automated systems could eventually help doctors to catch early-stage arthritis even when they aren’t specifically looking for it – and before people notice symptoms such as stiffness and pain in their knees or hips. Physiotherapy could stop early-stage arthritis from getting worse and there are several medications under development that could slow the disease’s progression.

Versatility of bone density (DEXA) scans

Bone density (DEXA) scans use X-rays at a much lower dose than standard X-ray images. Doctors routinely use them to check for osteoporosis, a condition that weakens the bones and makes them more likely to break. For osteoarthritis, which causes the protective cartilage on the ends of bones to break down, a standard X-ray is usually needed.

But DEXA scans can be as good as regular X-ray images when it comes to spotting signs of arthritis. Researchers used DEXA scans from around 40,000 UK Biobank participants to develop an AI algorithm that can spot hints of arthritis in these scans. It can identify how severe the disease is and pick out people who might need a full knee or hip replacement.

Early detection could transform care

“It’s about developing a more personalised approach for people with osteoarthritis.”

Professor Jonathan Tobias, University of Bristol, UK

This means routine osteoporosis checks could also look for signs of developing arthritis, which could get people treated earlier, and save the NHS time and money. “The idea is to pick out people with osteoarthritis who didn’t know they have it,” explains study team member Jonathan Tobias. Doctors could then prescribe physiotherapy, which can stop the condition from getting worse, allowing for more years living in better health.

There are also various medications under development that can slow down the disease’s progression – although so far, many of these new drugs have failed to make people better. “The concern is that that’s because they’re lumping all these different types of osteoarthritis together,” Tobias explains.

DEXA analysis could help to identify different types of arthritis. “That might enable you to find drugs that are effective in one particular type and not another,” Tobias says. “It’s about developing a more personalised approach for people with osteoarthritis.”

Header image credit: Rheumatology (Oxford), Volume 61, Issue 9, September 2022, Pages 3586–3595