AI interpretation of MRI images may provide more accurate RA prediction

Early inflammatory arthritis is often undifferentiated but can progress to established rheumatoid arthritis or another arthropathy. Alternatively, they may disappear spontaneously or remain undifferentiated indefinitely. Erosion is an important prognostic factor and can be detected with magnetic resonance imaging (MRI). In addition, MRI allows direct visualization and assessment of (tendon) synovitis and bone marrow edema.

Predicting early RA from MRI images of hands and feet could help people receive timely treatment and change the course of the disease. Traditionally, this has been done by radiologists and rheumatologists using scoring sheets to manually identify key features from MRI scans. But now, artificial intelligence (AI) interpretation of his MRI images may provide more accurate predictions than visual scoring by medical staff.

A summary presented by Lee et al. of Leiden University Medical Center describes how deep learning AI can automatically analyze scans for early prediction of rheumatoid arthritis in patients with clinically suspected joint pain. is explained in detail. The model was first trained to understand anatomy, then to distinguish between patients and healthy controls, and finally to find image features that predict the development of rheumatoid arthritis. AI analyzed hand and foot scans of 1,974 people with early-onset arthritis or clinically suspected joint pain, of whom 651 developed rheumatoid arthritis. Results from the test set conducted showed that this model can predict rheumatoid arthritis with accuracy approaching that achieved by human experts. This was useful for scanning hands or feet.

The authors conclude that AI interpretation of MRI scans has the potential to automatically predict RA. Further training of the model using MRI data from healthy controls would likely improve accuracy. Future studies will focus on predicting RA, particularly in undifferentiated arthritis, as a subgroup of the early-onset arthritis group. Moreover, this new method not only confirms the importance of known inflammatory features such as synovial inflammation, but also has the potential to identify new imaging biomarkers in the long term, the disease underlying early rheumatoid arthritis. Deepen your understanding of the process.