Researchers from the University of Jyväskylä and the Central Finland Health Care District have developed an AI-based neural network to detect early knee osteoarthritis using X-ray images. AI was able to match a doctor’s diagnosis 87% of the time. The result is important because radiographs are the primary diagnostic method for early knee osteoarthritis. Early diagnosis can save the patient from unnecessary investigations, treatments and even knee replacement surgery.
Osteoarthritis is the most common joint disease worldwide. In Finland alone, it causes up to 600,000 visits to the doctor every year. According to estimates, it costs the national economy up to 1 billion euros every year.
The new AI-based method was trained to use X-rays to detect a radiological feature that predicts osteoarthritis. The finding is currently not included in the diagnostic criteria, but is evaluated by orthopedists as an early sign of arthrosis. The method was developed in the Digital Health Intelligence Lab at the University of Jyväskylä as part of the AI Hub Central Finland project. It uses neural network technologies that are widely used around the world.
“The aim of the project was to train the AI to recognize an early sign of osteoarthritis on an X-ray. Something that experienced doctors can visually distinguish from the image, but cannot do automatically,” explains lead researcher Anri Patron, explaining the development of the method.
In practice, the AI tries to recognize whether there are spikes on the shin bones in the knee joint or not. Shin spikes can be a sign of osteoarthritis.
The reliability of the method was evaluated together with specialists from the Central Finland Healthcare District.
“Around 700 X-ray images were used to develop the AI model, after which the model was validated with around 200 X-ray images. of cases, which is a promising result,” describes Patron.
AI can support early detection of osteoarthritis in primary care
Lecturer Sami Äyrämö, head of the Digital Health Intelligence Laboratory at the University of Jyväskylä, explains that the development of AI models for diagnosing early osteoarthritis is active worldwide.
“Previously, several AI models were developed to detect knee osteoarthritis. These models can detect severe cases that would be easily detected by any specialist. However, the previously developed methods are not accurate enough to detect the early-stage manifestations. The method, which is now being developed, aims in particular at early detection using X-rays, for which there is a great need.”
The goal is for an AI to be able to use X-ray images to detect early signs of knee arthrosis in the future, so that the initial diagnosis can be made more often by general practitioners.
The project was carried out in cooperation with the Central Finnish Health District. Juha Paloneva, CEO of Central Finland Health District and professor of surgery, says osteoarthritis can be effectively treated in its early stages.
“If we can make the diagnosis at an early stage, we can avoid uncertainties and expensive tests like MRI scans. In addition, the patient can be motivated to take measures to slow down or even stop the progression of the symptomatic arthrosis. “In the best case, the patient could even avoid joint replacement surgery,” summarizes Paloneva.
