Abstract:
Objective: Characterising the morphological differences between healthy and early osteoarthritic (EOA) trapeziometacarpal (TMC) joints is important for understanding osteoarthritis onset, and early detection is important for treatment and disease management. This study has two aims: first, to characterise morphological differences between healthy and EOA TMC bones. The second aim was to determine the efficacy of using a statistical shape model (SSM) to detect early signs of osteoarthritis (OA). Methods: CT image data of TMC bones from 22 asymptomatic volunteers and 47 patients with EOA were obtained from an ongoing study and used to generate a SSM. A linear discriminant analysis (LDA) classifier was trained on the principal component (PC) weights to characterise features of each group. Multivariable statistical analysis was performed on the PC to investigate morphologic differences. Leave-one-out classification was performed to evaluate the classifiers performance. Results: We found that TMC bones of EOA subjects exhibited a lower aspect ratio (P = 0.042) compared with healthy subjects. The LDA classifier predicted that protrusions (up to 1.5 mm) at the volar beak of the first metacarpal were characteristic of EOA subjects. This was accompanied with widening of the articular surface, deepening of the articular surface, and protruding bone growths along the concave margin. These characteristics resulted in a leave-one-out classification accuracy of 73.9% (95% CI [61.9%, 83.8%]), sensitivity of 89.4%, specificity of 40.9%, and precision of 75.9%. Conclusion: Our findings indicate that morphological degeneration is well underway in the EOA TMC joint, and shows promise for a clinical tool that can detect these features automatically.