Computational protocol: Cortical and trabecular morphology is altered in the limb bones of mice artificially selected for faster skeletal growth

Similar protocols

Protocol publication

[…] Following the whole body scan, the left tibia was dissected at the knee joint, soft tissues were removed, and the bone was scanned on an Xradia Versa 520 μCT scanner (Carl Zeiss Inc., Thornwood, NY) at an isotropic voxel size of 1.9 μm (voltage: 80 kV, power: 7 W, exposure time: 0.95 s). Three individuals (1× LS1, 2× LS2), were not recovered following the SkyScan whole body scan. A total of 994 slices were captured from each scan beginning at the tibia-fibula junction and proceeding distally. Scans were then converted to image stacks using XMReconstructor (Xradia; Carl Zeiss Inc., Thornwood, NY).For each sample, a sub-region equivalent to 1 mm (526 slices) of cortical bone, starting 0.5 mm distal to the tibia-fibula junction, was imported into Fiji (ImageJ v1.50 e) for processing using a custom-written macro, as follows: Stacks were first converted to binary images using a global threshold. Next, the “Analyze Particles” function in Fiji was used to digitally fill open spaces in the bone under 1000 μm3 (~145 voxels), representing osteocyte lacunae and other noise, leaving a stack containing larger spaces representing canals, . This stack was duplicated, and the function was run once more to fill the canals, leaving a solid block in which all spaces within the cortex had been filled. Bone volumes of the two stacks were computed, and the ratio of the smaller (with canals) to larger (filled canals) volume was used to calculate cortical porosity in %, i.e., as (1-volume with canals/volume without canals) * 100%.Canal thickness (in mm) was derived from the stack with the canals present, by first removing the cortex and the background, leaving only the canals as 3D objects, and running the “Thickness” function in the BoneJ plugin. To obtain canal orientation, the 3D canal stack was processed using the “Skeletonize (2D/3D)” function, followed by the “AnalyzeSkeleton” plugin in Fiji. The output of these functions comprises canal branch information with lengths and 3D coordinates of the ends of the branch. These data were exported to Matlab 2016 (Mathworks, Natick, MA), where a custom script was used to remove branches under 100 μm in length. We calculated the orientation of the canals relative to the longitudinal axis passing through the centroids of each slice, as described, .Finally, cortical cross-sectional geometry variables were obtained from the transverse section at the middle of the filled stack (i.e., from a single slice 1 mm distal to the tibia-fibula junction), using the “Slice Geometry” function in the BoneJ plugin. We obtained cortical area (Ct.Ar, in mm2), mean cortical thickness (Ct.Th, in mm), the moment of inertia about the principal axes (Imax, mm4, Imin, mm4), and the polar section modulus (i.e. ZP , in mm3). ZP is frequently used as a measure of long bone bending strength at the midshaft of long bones, , . [...] Linear mixed models were used to analyse time-dependent changes in bone indices (i.e., repeated measures) within the lines and treatment groups. We used a factorial model in which individual was treated as a random factor, while time and treatment (OVX vs Sham surgery) were considered fixed factors (Statistica, v12.0, StatSoft Inc. Tulsa, OK). Statistical significance of differences between the treatments within a line at each time point, and within treatments over time, was determined using a Fisher’s least square difference (LSD) post-hoc test. […]

Pipeline specifications

Software tools ImageJ, BoneJ, AnalyzeSkeleton, Statistica
Applications Miscellaneous, Microscopic phenotype analysis
Organisms Mus musculus
Diseases Bone Diseases, Alveolar Bone Loss
Chemicals Estrogens