A Longitudinal Low Dose μCT Analysis of Bone Healing in Mice: A Pilot Study

Figure 2

Interobserver agreement in the determination of 3D ROI. A Bland-Altman plot was used to graph differences in the absolute number of voxels per 3D ROI as a function of the average size of the defect. This made it possible to assess the agreement between three observers in the determination of the 3D ROI. Eighteen µCT images from (a) unreamed femurs with a defect and (b) reamed femurs with a defect were blindly analyzed. The two-by-two comparison of the three observers (A-B, B-C, and C-A) showed a 95% confidence interval from −226.5 to +226.4 and from −245.9 to +247.1 for unreamed femurs with a defect and reamed femurs with a defect, respectively. An interclass correlation (ICC) was used to compare interobserver variability. The ICC scores were 0.864 (very good, ) and 0.905 (ideal, ) for the unreamed femurs with a defect and the reamed femurs with a defect, respectively.

(a)

(b)