Development of a 3D finite element model of trabecular bone: quantitative validation against 3D full-field strain measurements

Osteoporosis is the most common chronic metabolic bone disease related to various factors including menopause and aging. Since bone turnover is increased, the metabolic activity changes. These alterations lead to net-loss of bone tissue, with decreased structural connectivity and apparent density, which subsequently results in a weaker structure. Fragility fractures are the clinical outcome of osteoporosis. Bone mineral density (BMD) measurement or the development of algorithms such as Fracture Risk Assessment Tool Model are two ways to estimate the fracture risk, but their predictive capabilities are still limited. Hence, a more detailed evaluation of pathological risk of fracture is needed. In the present thesis, I aimed to improve predictions of bone damage and fracture. This was done by the development of a subject-specific microstructural finite element (µFE) model based on high-resolution images from micro-computed tomography (µCT). Moreover, a procedure to validate the model by using experimental measurements from Digital Volume Correlation (DVC) was implemented. Homogenous, isotropic, and linear elastic µFE model was generated with two different sets of boundary conditions (BCs). DVC experimental measurements at three different levels of resolution were analysed by a procedure based on interpolation and directly compared to predicted values by µFE model. The implemented method could take the first step towards validating the µFE models. The analysis showed that volumetric strain maps from DVC are more influenced by noise effects than are displacement maps. This influence was more visible at high-resolution level with respect to downscaled data. Comparisons between µFE and DVC showed that strain fields differ significantly between the two methods and that the µFE model can detect only high volumetric strain regions where cracks are about to occur. With regards to the displacement field, µFE model can be accurate in two directions. The results suggest further development in methods for both the DVC and the µFE model, for instance adding element specific material properties based on BMD values to the model or improving the DVC scanning techniques could represent some possibilities.