Comparability of CT and μCT-Extracted Femoral Diaphyseal Data in Primates

The increased use of medical CT and μCT for internal analyses of primate bones has led to studies combining both data types to maximize sample sizes. However, comparability of data between modalities has yet to be demonstrated. Here, we test whether CT and μCT data from the primate femoral diaphysis can be confidently combined.

Femora from 17 specimens (Pan, Gorilla, Pongo, Hylobates, Ateles, Lagothrix, Mandrillus, Pithecia) were scanned using both CT modalities. Six segmentation methods were evaluated: Manual, 1/3 and 2/3 of maximum value, Full-Width-at-Half-Maximum, Kohler method, and Buie method in Dragonfly 3D World. The R package 'morphomap' was used to extract cortical bone thicknesses and calculate cross-sectional geometric (CSG) properties in cross sections from 80% (proximal) to 20% (distal) of femoral biomechanical length. Percent error was calculated by property and location. Linear regressions evaluated body mass as a predictor of error, and correction factors were developed using regression coefficients and location-specific mean errors.

Medical CT consistently underestimated most properties relative to μCT, while cortical thickness and CA were generally overestimated. The greatest percent errors occurred at distal and proximal diaphyseal locations in most comparisons. Regressions showed significant associations between body mass and error at distal sites. Application of regression-based and mean percent corrections reduced error to acceptable ranges.

Body size and segmentation method significantly impact comparability between medical CT and μCT in cortical thickness and CSG data. Applying location- and property-specific corrections improves measurement reliability in mixed-modality datasets and enhances comparisons across taxa.