Fully automatic quantification of knee osteoarthritis severity on plain radiographs.

OBJECTIVE

Although knee osteoarthritis (OA) is a major public health issue causing chronic disability, there is no objective or accurate method for measurement of the structural severity in general clinical practice. Here we have established a fully automatic program KOACAD (knee OA computer-aided diagnosis) to quantify the major OA parameters on plain knee radiographs, validated the reproducibility and reliability, and investigated the association of the parameters with knee pain.

METHODS

KOACAD was programmed to measure joint space narrowing at medial and lateral sides, osteophyte formation, and joint angulation. Anteroposterior radiographs of 1979 knees of a large-scale cohort population were analyzed by KOACAD and conventional categorical grading systems.

RESULTS

KOACAD automatically measured all parameters in less than 1s without intra- or interobserver variability. All parameters, especially medial joint space narrowing, were significantly correlated with the conventional gradings. In the parameters, osteophyte formation was associated with none of the joint space parameters, suggesting different etiologic mechanisms between them. Multivariate logistic regression analysis after adjustment for age and confounding factors revealed that medial joint space narrowing and varus angulation of knee joints were risk factors for the presence of pain (594/1979 knees), while neither lateral joint space nor osteophyte area was.

CONCLUSION

KOACAD was shown to be useful for objective, accurate, simple and easy evaluation of the radiographic knee OA severity in daily clinical practice. This system may also serve as a surrogate measure for the development of disease-modifying drugs for OA, just as bone mineral density does in osteoporosis.