Field test of the reproducibility of automated measurements of medial tibiofemoral joint space width derived from standardized knee radiographs.

  • 1 June 1999
    • journal article
    • Vol. 26  (6) , 1359-65
Abstract
To estimate the reproducibility of computerized measurements of minimum joint space width (JSW) in the medial tibiofemoral compartment in knee radiographs (semiflexed AP view) obtained from clinical radiology units. Technologists from 5 clinical radiology units were trained in the performance of the fluoroscopically assisted semiflexed AP knee examination. Each of 44 subjects (34 with knee osteoarthritis, OA, 10 with bilaterally normal knees) were examined within 7 days in 2 of the 5 units. The examination in each unit was repeated 1 week later. Minimum JSW was measured on digitized radiographic images with computer software that corrected for radiographic magnification. Despite ongoing quality control by technologists, 11% of radiographs were flawed with respect to the protocol standard for knee rotation and 36% with respect to the standard for knee flexion. The standard error of measurement (SEm) of JSW in 174 knees that were examined twice in the same unit was 0.32 mm (SEm = 0.25 mm for the subset of 76 paired radiographs with uniformly high quality). The overall between-unit SEm was 0.45 mm. Within-unit, but not between-unit, precision was related to the technical quality of the radiographs. Precision was unrelated to subject age, sex, race, weight, and radiographic severity of knee OA. The within-unit precision of JSW measurements from all pairs of semiflexed views (irrespective of technical quality) represented a notable improvement over that observed in radiographs with flawed knee rotation or flexion (as would be the case in conventional extended knee views). In future applications of this technique, assurance of technical quality by an independent observer should result in a level of measurement precision that will permit the design of clinical trials of disease modifying OA drugs with fewer subjects and/or shorter duration of treatment than is possible with conventional knee radiography.

This publication has 0 references indexed in Scilit: