Purpose: The purpose of this study was to evaluate the inter- and intra-observer variability of the computerized radiologic measurements using Keops(®) and to determine the bias between the software and the standard paper measurement.
Methods: Four individuals measured all frontal and sagittal variables on the 30 X-rays randomly selected on two occasions (test and retest conditions). The Bland-Altman plot was used to determine the degree of agreement between the measurement on paper X-ray and the measurement using Keops(®) for all reviewers and for the two measures; the intraclass correlation coefficient (ICC) was calculated for each pair of analyses to assess interobserver reproducibility among the four reviewers for the same patient using either paper X-ray or Keops(®) measurement and finally, concordance correlation coefficient (rc) was calculated to assess intraobserver repeatability among the same reviewer for one patient between the two measure using the same method (paper or Keops(®)).
Results: The mean difference calculated between the two methods was minimal at -0, 4° ± 3.41° [-7.1; 6.4] for frontal measurement and 0.1° ± 3.52° [-6.7; 6.8] for sagittal measurement. Keops(®) has a better interobserver reproducibility than paper measurement for determination of the sagittal pelvic parameter (ICC = 0.9960 vs. 0.9931; p = 0.0001). It has a better intraobserver repeatability than paper for determination of Cobbs angle (rc = 0.9872 vs. 0.9808; p < 0.0001) and for pelvic parameter (rc = 0.9981 vs. 0.9953; p < 0.0001).
Conclusions: We conclude that Keops(®) has no bias compared to the traditionally paper measurement, and moreover, the repeatability and the reproducibility of measurements with this method is much better than with similar standard radiologic measures done manually in both frontal and sagittal plane and that the use of this software can be recommended for clinical application.
Level of evidence: Diagnostic, level III.