Objective: Herewith, we report the development of Orthopedic Digital Image Analysis (ODIA) software that is developed to obtain quantitative measurements of knee osteoarthritis (OA) radiographs automatically. Manual segmentation and measurement of OA parameters currently hamper large-cohort analyses, and therefore, automated and reproducible methods are a valuable addition in OA research. This study aims to test the automated ODIA measurements and compare them with available manual Knee Imaging Digital Analysis (KIDA) measurements as comparison. Design: This study included data from the CHECK (Cohort Hip and Cohort Knee) initiative, a prospective multicentre cohort study in the Netherlands with 1,002 participants. Knee radiographs obtained at baseline of the CHECK cohort were included and mean medial/lateral joint space width (JSW), minimal JSW, joint line convergence angle (JLCA), eminence heights, and subchondral bone intensities were compared between ODIA and KIDA. Results: Of the potential 2,004 radiographs, 1,743 were included for analyses. Poor intraclass correlation coefficients (ICCs) were reported for the JLCA (0.422) and minimal JSW (0.299). The mean medial and lateral JSW, eminence height, and subchondral bone intensities reported a moderate to good ICC (0.7 or higher). Discrepancies in JLCA and minimal JSW between the 2 methods were mostly a problem in the lateral tibia plateau. Conclusions: The current ODIA tool provides important measurements of OA parameters in an automated manner from standard radiographs of the knee. Given the automated and computerized methodology that has very high reproducibility, ODIA is suitable for large epidemiological cohorts with various follow-up time points to investigate structural progression, such as CHECK or the Osteoarthritis Initiative (OAI).

Objective: To summarize available evidence on the association between hip shape as quantified by statistical shape modeling (SSM) and the incidence or progression of hip osteoarthritis. Design: We conducted a systematic search of five electronic databases, based on a registered protocol (available: PROSPERO CRD42020145411). Articles presenting original data on the longitudinal relationship between radiographic hip shape (quantified by SSM) and hip OA were eligible. Quantitative meta-analysis was precluded because of the use of different SSM models across studies. We used the Newcastle–Ottawa Scale (NOS) for risk of bias assessment. Results: Nine studies (6,483 hips analyzed with SSM) were included in this review. The SSM models used to describe hip shape ranged from 16 points on the femoral head to 85 points on the proximal femur and hemipelvis. Multiple hip shape features and combinations thereof were associated with incident or progressive hip OA. Shape variants that seemed to be consistently associated with hip OA across studies were acetabular dysplasia, cam morphology, and deviations in acetabular version (either excessive anteversion or retroversion). Conclusions: Various radiographic, SSM-defined hip shape features are associated with hip OA. Some hip shape features only seem to increase the risk for hip OA when combined together. The heterogeneity of the used SSM models across studies precludes the estimation of pooled effect sizes. Further studies using the same SSM model and definition of hip OA are needed to allow for the comparison of outcomes across studies, and to validate the found associations.

Objective: To design an automated workflow for hip radiographs focused on joint shape and tests its prognostic value for future hip osteoarthritis. Design: We used baseline and 8-year follow-up data from 1,002 participants of the CHECK-study. The primary outcome was definite radiographic hip osteoarthritis (rHOA) (Kellgren–Lawrence grade ≥2 or joint replacement) at 8-year follow-up. We designed a method to automatically segment the hip joint from radiographs. Subsequently, we applied machine learning algorithms (elastic net with automated parameter optimization) to provide the Shape-Score, a single value describing the risk for future rHOA based solely on joint shape. We built and internally validated prediction models using baseline demographics, physical examination, and radiologists scores and tested the added prognostic value of the Shape-Score using Area-Under-the-Curve (AUC). Missing data was imputed by multiple imputation by chained equations. Only hips with pain in the corresponding leg were included. Results: 84% were female, mean age was 56 (±5.1) years, mean BMI 26.3 (±4.2). Of 1,044 hips with pain at baseline and complete follow-up, 143 showed radiographic osteoarthritis and 42 were replaced. 91.5% of the hips had follow-up data available. The Shape-Score was a significant predictor of rHOA (odds ratio per decimal increase 5.21, 95%-CI (3.74–7.24)). The prediction model using demographics, physical examination, and radiologists scores demonstrated an AUC of 0.795, 95%-CI (0.757–0.834). After addition of the Shape-Score the AUC rose to 0.864, 95%-CI (0.833–0.895). Conclusions: Our Shape-Score, automatically derived from radiographs using a novel machine learning workflow, may strongly improve risk prediction in hip osteoarthritis.