The effects of geometric and threshold definitions on cortical bone metrics assessed by in vivo high-resolution peripheral quantitative computed tomography.

This study evaluates in vivo methods for calculating cortical thickness (Ct.Th) with respect to sensitivity to tissue-level changes in mineralization and the ability to predict whole-bone mechanical properties. Distal radial and tibial images obtained from normal volunteers using high-resolution peripheral quantitative computed tomography (HR-pQCT) were segmented using three thresholds including the manufacturer default and +/-5% in terms of equivalent mineral density. Ct.Th was determined in two ways: using a direct three-dimensional (3D) method and using an annular method, where cortical bone volume is divided by periosteal surface area. D(comp) (mg HA/cm(3)) was calculated based on the mean density-calibrated linear attenuation values of the cortical compartment. Finite element analysis was performed to evaluate the predictive ability of the annular and direct Ct.Th methods. Using the direct 3D method, a +/-5% change in threshold resulted in a 2% mean difference in Ct.Th for both the radius and tibia. An average difference of 5% was found using the annular method. The change in threshold produced changes in D(comp) ranging 0.50-1.56% for both the tibia and radius. Annular Ct.Th correlated more strongly with whole-bone apparent modulus (R(2)=0.64 vs. R(2)=0.41). Both thickness calculation methods and threshold selection have a direct impact on cortical parameters derived from HR-pQCT images. Indirectly, these results suggest that moderate changes in tissue-level mineralization can affect cortical measures. Furthermore, while the direct 3D Ct.Th method is less sensitive to threshold effects, both methods are moderate predictors of mechanical strength, with the annular method being the stronger correlate.