A novel framework for assessing variations in 3D geometry of mouse tibiae in longitudinal preclinical studies

The mouse tibia is commonly used for testing osteoporosis treatments preclinically. In vivo micro-Computed Tomography (microCT) is the standard for monitoring longitudinal bone changes. Standard morphometry applied on images provides temporal changes in scalar geometric properties. However, these properties lack precision due to their limited dimensionality and inability to describe non-uniform 3D geometric changes. This PhD project aimed to develop a novel statistical framework enabling comprehensive assessment of spatio-temporal bone changes. The proposed framework combined longitudinal microCT imaging, image processing, Principal Component Analysis (PCA) and post-processing statistical analysis. First, the accurate, robust PCA-based model was developed, with less than one voxel error in describing bone shapes. Initially, the framework was applied to longitudinal microCT images of the tibial midshaft in an osteoporotic mouse model, elucidating 3D changes induced by in vivo mechanical loading (ML) including bone turnover at the anterior crest (0.103 mm) and posterior-lateral compartment. Next, it was applied to the entire cortex of osteoporotic mice, which received ML, Parathyroid hormone (PTH), and a combination of both (PTHML). The model revealed high bone turnover proximally, with posterior thickening, medial expansion and anterior thickening, highlighting increased anabolic effects of the PTHML. The first modes detected axial shape variations, coupled with positional misalignments. The latter was up to 0.142 mm in the axial position between the two ages in the diseased group. Motivated by this, the next study demonstrated non-negligible instance alignment effects in the shape analysis, suggesting a registration protocol that allowed meaningful biological inferences. Finally, a novel Partial Least Squares regression model revealed that bone length increase, posterior thickening and area expansion of the diaphysis described increased bone strength due to combined PTH and ML (R2=81%). This novel statistical framework offers a precise strategy for assessing treatment efficacy by measuring 3D bone geometry and strength changes in preclinical studies.