Micro level finite element models of bone have been extensively used in the literature to examine its mechanical behaviour and response to loads. Techniques used previously to create these models involved CT attenuations or images (e.g. micro-CT, MRI) of real bone samples. The computational models created using these methods could only represent the samples used in their construction and any possible variations due to factors such as anatomical site, sex, age or degree of osteopo-rosity cannot be included without additional sample collection and processing. This study considers the creation of virtual finite element models of trabecular bone, i.e. models that look like and mechanically behave like real trabecular bone, but are generated computationally.

The trabecular bone is anisotropic both in terms of its micro-architecture and its mechanical properties. Considerable research shows that the key determinants of the mechanical properties of bone are related to its micro-architecture. Previous studies have correlated the apparent level mechanical properties with bone mineral density (BMD), which has also been the principal means of diagnosis of osteoporosis. However, BMD alone is not sufficient to describe bone micro-architecture or its mechanical behaviour. This study uses a novel approach that employs BMD in conjunction with micro-architectural indices such as trabecular thickness, trabecular spacing and degree of anisotropy, to generate virtual micro-architectural finite element models. The approach permits generation of several models, with suitable porous structure, for the same or different levels of osteoporosity. A series of compression and shear tests are conducted, numerically, to evaluate the apparent level orthotropic elastic properties. These tests show that models generated using identical micro-architectural parameters have similar apparent level properties, thus validating this initial bone modelling algorithm. Numerical tests also clearly illustrate that poor trabecular connectivity leads to inferior mechanical behaviour even in cases where the BMD values are relatively high. The generated virtual models have a range of applications such as understanding the fracture behaviour of osteoporotic bone and examining the interaction between bone and implants.