Deep learning-measured effusion-synovitis volume change on MRI linked to knee osteoarthritis progression

Longitudinal change in deep learning-based effusion-synovitis volume (ΔESV) over 48 months was associated with concurrent knee OA progression (p<0.001). ΔESV showed stronger associations for ΔKL grade than semiquantitative MOAKS (ΔIDD=0.22, p=0.002).

• Retrospective analysis of 2,469 MRIs from the Osteoarthritis Initiative; deep learning segmented effusion-synovitis volume.
• ΔESV also showed stronger associations for WORMS total and meniscus subscales vs. MOAKS, but not for BMELL, cartilage, or symptoms.
• Limitation: single-cohort, no external validation; predictive utility remains unproven.

RadPigeon summaries are original and for information only. They are not clinical advice.