Objective: This study was conducted 1) to determine the feasibility of using computer programs to measure radiographic joint space width and estimate erosion volume in the hands of patients with rheumatoid arthritis (RA), and 2) to compare the new computer-based methods with established scoring methods.
Methods: To measure the joint space width in the finger and wrist joints of RA patients, hand and wrist radiographic films were scanned using a tabletop scanner and analyzed with programs written using the "macro" capabilities of NIH Image software. Estimation of erosion volume was determined by utilizing gray-scale intensity to calibrate bone density units per mm3, which made possible comparisons between the erosions and noneroded, anatomically similar sites.
Results: In 3 sets of duplicate measurements of joint space width on 79, 48, and 48 finger and wrist joints, the mean absolute deviation from the mean of the 2 measurements was 0.036 mm (SD 0.034), 0.032 mm (SD 0.049), and 0.021 mm (SD 0.016), respectively. Joint space measurements and scoring of joint space narrowing both demonstrated a difference between active treatment and placebo in an old trial set on gold therapy (P = 0.03 and P = 0.01, respectively). Two repeated measurements of bone density units in the bones of 3 different hands differed from the mean of the measurement by 2.29-4.04%. In 2 experiments, estimates of erosion volume showed a greater difference between gold therapy and placebo than did erosion scores in the trial set (P = 0.049 and P = 0.016 versus P = 0.27).
Conclusion: Computer-based methods for measuring finger and wrist joint spaces and estimating erosion volume in patients with RA agree with the results of radiographic scoring of erosions and joint space narrowing.