Diagnosis of Osteoarthritis by Cartilage Surface Smoothness Quantified Automatically from Knee MRI

OBJECTIVE We investigated whether surface smoothness of articular cartilage in the medial tibiofemoral compartment quantified from magnetic resonance imaging (MRI) could be appropriate as a diagnostic marker of osteoarthritis (OA). METHOD At baseline, 159 community-based subjects aged 21 to 81 with normal or OA-affected knees were recruited to provide a broad range of OA states. Smoothness was quantified using an automatic framework from low-field MRI in the tibial, femoral, and femoral subcompartments. Diagnostic ability of smoothness was evaluated by comparison with conventional OA markers, specifically cartilage volume from MRI, joint space width (JSW) from radiographs, and pain scores. RESULTS A total of 140 subjects concluded the 21-month study. Cartilage smoothness provided diagnostic ability in all compartments (P < 0.0001). The diagnostic smoothness markers performed at least similar to JSW and were superior to volume markers (e.g., the AUC for femoral smoothness of 0.80 was higher than the 0.57 for volume, P < 0.0001, and marginally higher than 0.73 for JSW, P = 0.25). The smoothness markers allowed diagnostic detection of pain presence (P < 0.05) and showed some correlation with pain severity (e.g., r = -0.32). The longitudinal change in smoothness was correlated with cartilage loss (r up to 0.60, P < 0.0001 in all compartments). CONCLUSIONS This study demonstrated the potential of cartilage smoothness markers for diagnosis of moderate radiographic OA. Furthermore, correlations between smoothness and pain values and smoothness loss and cartilage loss supported a link to progression of OA. Thereby, smoothness markers may allow detection and monitoring of OA-supplemented currently accepted markers.