Osteoporosis is a major metabolic bone disease that causes reduced bone mass, deteriorated bone microstructural and increased fracture risk. In clinical practice, the gold standard to examine bone quality and evaluate fracture risk is using dual energy X-ray absorptiometry through measurements of areal bone mineral density (aBMD). However, it has been well accepted that in addition to aBMD, bone geometry, microstructure and material properties also play important roles in determining overall bone mechanical competence, which is directly related to fracture risk. High-resolution peripheral quantitative computed tomography (HR-pQCT) has the capability to image three-dimensional (3D) bone microstructures in vivo and provide quantitative measurements of bone mineral density as well as cortical and trabecular microstructure. Based on the HR-pQCT images, micro finite element (µFE) models can be constructed to directly estimate bone strength. HR-pQCT has become a widely used imaging tool in clinical research to evaluate the effect of aging, drug treatment, and metabolic bone disease on bone quality. The work in this thesis focuses on evaluating the accuracy and capability of HR-pQCT in quantifying microstructural properties of human radius and tibia bone, exploring its prediction power of whole bone strength and discussing potential applications in clinical studies.

In this thesis, we quantified the accuracy of the standard HR-pQCT microstructural measurements of human distal radius and tibia through comparisons with gold standard µCT-based morphological measures. The results showed that the BV/TV^d , Tb.N^* , Tb.Th and Tb.Sp from HR-pQCT were significantly and highly correlated with those from gold-standard µCT measurements. Strong correlations between the HR-pQCT µFE predictions and direct mechanical testing measures suggest that HR-pQCT µFE is a robust method to determine bone mechanical properties.

In a clinical setting, standard HR-pQCT scans are performed on the non-dominant wrist (usually the left) and the corresponding tibia. However, the contralateral side is selected for scanned when there is a fracture the non-dominant wrist. It remains unclear whether the dominant side is representative of the non-dominant side and how much error it will bring into a study where subjects include mixed scans of both sides. In this thesis, we applied HR-pQCT and µCT based morphological and mechanical measurements to characterize the symmetric nature of distal radius and tibia. We found that the right radius tend to be larger than the left radius. However, at the tibia, the bone size was found to be similar between left and right. By micro computed tomography (µCT), microstructural parameters such as BV/TV were also found to be larger at the right radius, while no difference was found at the tibia. Trabecular number, trabecular thickness, trabecular separation and cortical thickness were not different between left and right radius. µFE analyses demonstrated that stiffness and strength of right radius were significantly higher than left radius, while there was no difference at the tibia.

The standard clinical region of interest HR-pQCT is recommended by the manufacturer;​ however, it is not clear whether a segment HR-pQCT scan is representative of whole bone mechanical properties. Therefore we quantified the associations of microstructural and mechanical measurements of the radius and tibia segments with whole bone stiffness and examined if we can improve the correlation when we select a different region. The microstructural and mechanical measurements at the two regions next to the standard HR-pQCT segment (proximal and distal) were also examined. The results showed that the bone microstructure from proximal and distal sections is highly correlated to standard region at both distal radius and tibia. The mechanical properties of the three segments were strongly correlated with overall bone mechanical properties. The microstructural measurements at the most distal section were correlated with whole bone stiffness better compared to those from standard and proximal regions.

DXA is incapable of discriminating patients with wrist fracture from those without. In this study, we examined the microstructural and mechanical properties in patients with and without wrist fracture through HR-pQCT based analyses. We demonstrated that wrist fracture patients had lower plate and rod bone volume fraction, less plate and rod trabecular number, thinner cortex and lower whole bone stiffness and strength, compared to healthy controls. Failure analyses also depicted significantly lower trabecular plate compression and tension failure fraction in wrist fracture patients