The femurs were scanned using an isotropic voxel size of 19 µm (SkyScan 1172, v. 1.5; SkyScan, Aarteselar, Belgium) using energy settings of 50 kV and 200 µA, a 0.5-mm aluminum filter, and 8 repeated scans. Image reconstruction was performed (SkyScan NRecon package v. 1.5.1.4) by correcting for ring artifacts and beam hardening (20%). Following reconstruction, the individual fracture lines were identified by simultaneously viewing multiple orthogonal slices (Skyscan DataViewer v. 1.4). The region of interest for each bone was determined as being approximately 3 mm proximal and distal to the fracture line (150 images) (Nyman et al. 2009 (link)). Within that region of interest (ROI), semi-automatic segmentation was used on each 2D image to identify the circumferential boundaries of the calluses (Figure 1) (Matlab v. 7.6.0; Mathworks Inc.; SkyScan CTAn v. 1.9.1.0). Calibration of bone mineral density (BMD) was carried out according to the system manufacturer’s protocol. A water phantom and 2 hydroxyapatite phantoms of known density (0.25 and 0.75 g/cm3) were scanned. To distinguish fully mineralized tissue from poorly mineralized tissue and soft tissue, 2 thresholds were used. Fully mineralized tissue was assumed to have a BMD of more than 0.642 g/cm3 (Morgan et al. 2009 (link)), resulting in grayscale values of 98–255. Poorly mineralized tissue was assumed to have a BMD value of between 0.410 and 0.642 g/cm3 (Isaksson et al. 2009 (link)), resulting in grayscale values of 68–97. The threshold values were chosen based on visual inspection of the images, qualitative comparison with histological sections, and previous studies. The following parameters were calculated from the callus region of interest for each specimen: total callus volume (TVc), fully mineralized bone volume (BVhigh), poorly mineralized tissue volume (BVlow), bone volume fraction (BVhigh / TVc), and average tissue mineral density (TMD). TMD was calculated by using only the voxels that exceeded the threshold for fully mineralized bone.