ABSTRACT

Quantification of the radiographic joint space width (JSW) of the knee is a widely used method to evaluate progression of osteoarthritis (OA) and remains the only structural outcome accepted by the USA FDA to demonstrate OA disease modification. The aim of this study was to evaluate the short and long term reproducibility of medial knee JSW measurements using Image J software (US NIH, Bethesda, MD, https://imagej.nih.gov/ij/) in medial knee OA.

Thirty-seven patients with symptomatic medial knee OA (Kellgren-Lawrence grade 2-3, pain on ambulation >30 mm on a 100-mm Visual Analog scale) were assessed. Semi-flexed (20°) fluoroscopy-guided posteroanterior knee radiographs (Lyon Schuss view) were obtained. Medial knee JSWs, defined as the narrowest inter-bone distance, were measured independently by two observers, one of which was an experienced observer and the other was a first time Image J user. Inter-observer variability (before and after one-hour training) and intra-observer variability, at short term (3 weeks) and long term (4 years) were calculated by using coefficient of variation (CV) and intra-class correlation coefficient (ICC).

One-hour training resulted in significant improvement in inter-observer variability (CV 6.6% vs 4.4%; ICC 0.89 vs0.98). Intraobserver variability at short term (3 weeks) was similar to long term (4 years) for the experienced observer (CV 3.8% vs 3.5%; ICC 0.97 vs 0.98). After one-hour training, intra-observer variability was also low for the first time Image J user (CV 3.8%; ICC 0.98).

Radiographic quantification of the JSW of the knee in patients with OA using Image J is reproducible. ImageJ, which is an open access and free software, is convenient to apply and requires only brief training to obtain high reproducibility.