Verification and Validation of MicroCT-based Finite Element Models of Bone Tissue Biomechanics

Non-destructive 3D micro-computed tomography (microCT) based finite element (microFE) model is popular in estimating bone mechanical properties in recent decades. From a fundamental scientific perspective, as the primary function of the skeleton is mechanical in nature, a lot of related biological and physiological mechanisms are mechano-regulated that becomes evident at the tissue scale. In all these research it is essential to known with the best possible accuracy the displacements, stresses, and strains induced by given loads in the bone tissue. Correspondingly, verification and validation of the microFE model has become crucial in evaluating the quality of its predictions. Because of the complex geometry of cancellous bone tissue, only a few studies have investigated the local convergence behaviour of such models and post-yield behaviour has not been reported. Moreover, the validation of their prediction of local properties remains challenging. Recent technique of digital volume correlation (DVC) combined with microCT images can measure internal displacements and deformation of bone specimen and therefore is able to provide experimental data for validation. However, the strain error of this experimental method tends to be a lot higher (in the order of several thousand microstrains) for spatial resolutions of 10-20 µm, typical element size of microFE models. Strictly speaking no validation of strain is possible. Therefore, the goal of this thesis it to conduct a local convergence study of cancellous bone microFE models generated using three microCT-based tissue modelling methods (homogeneous tetrahedral model, homogeneous hexahedral model and heterogeneous hexahedral model); to validate these models’ prediction in terms of displacement using the novel DVC technique; and finally to compare the strain field predicted by three tissue modelling methods, in order to explore the effect of specific idealisations/simplifications on the prediction of strain.