Automatic quantification of osteoarthritis in hand radiographs: validation of a new method to measure joint space width 1

SummaryObjectiveTo quantify osteoarthritis (OA) by automatic measurement of the joint space width (JSW) in the finger joints in hand radiographs, and validate this against semi-quantitative scoring.MethodsForty subjects with primary OA at multiple sites were selected, 20 of which had no joint space narrowing (JSN) in the hand, and 20 had moderate to severe JSN. Conventional hand radiographs were acquired and the image set was split into a training and test set, for developing and evaluating the method, respectively. No methods to quantify the JSW automatically in OA exist; therefore a method was developed. First the semi-automatic image analysis program identifies the metacarpal and interphalangeal joints. Subsequently, the corresponding joint margins and measurement interval are determined. Finally the JSW is measured in millimetres as the average distance between the joint margins enclosed by the measurement interval. The success rates of different steps of the image analysis were evaluated and a comparison was made between the JSW and the semi-quantitative Osteoarthritis Research Society International score for JSN.ResultsThe success rates for the identification of the joint locations and margins were 97.5% and 64.2%, respectively. The JSW decreased with increasing semi-quantitative scores, but increased as the joint was severely damaged. The agreement between the JSW and the semi-quantitative score was confirmed by a significant correlation (R2 = 0.54 and P < 0.01) while assessing hand pairs instead of individual joints.ConclusionThe JSW measurement closely reflects semi-quantitative scoring of JSN. Therefore, this method is expected to offer a reproducible and accurate measurement of the JSW in OA.