Osteoporosis is a skeletal disease which dramatically increases mortality and morbidity. In Europe the total direct cost for osteoporotic fractures are expected to be more than EUR76 billions in 2050. The current method to estimate the risk of fracture and therefore to decide which patients should be treated against bone loss is based on bone mineral density (BMD) analysis of the hip or of the spine, by means of dual energy X-rays absorptiometry (DXA) or quantitative computed tomography (QCT). However, such techniques have been found not to be reliable in predicting bone strength in in vitro studies. In the last decades the finite element (FE) method has been extensively used to try to enhance the prediction of bone strength in vitro. Nevertheless, such models should be meticulously validated through reliable experiments in vitro to evaluate to which extent they can accurately predict the reality. Therefore, the goal of this thesis was to apply QCT-based FE models of the human vertebra and femur, validate them versus accurate experiments performed in vitro on a large number of specimens and compare their predictive ability with the ones of densitometric measurements usually used in clinical applications.
The first study presented a methodology to compute bone volume fraction (BV/TV) from QCT BMD which could be applied for both vertebra and femur and used to define the material properties of the FE models. The second and third studies reported the results for the human vertebra, while the fourth and fifth studies reported the ones for the proximal femora. In particular, for both anatomical sites novel testing setups were designed to generate fractures which are usually observed in clinics on 37 vertebral bodies and on 72 femora as well as to compute their mechanical properties at the organ level. Moreover, these studies presented the developed automatic procedure to generate the specimen specific nonlinear homogenized voxel FE (hvFE) models from the QCT scans and the procedures to evaluate volumetric/areal BMD from QCT or DXA. The results of the Thesis showed that 1) similar calibration laws for both anatomical sites can be used to relate QCT BMD to BV/TV, 2) the hvFE models are better predictors of the vertebral and femoral mechanical properties than standard densitometric measurements, 3) and provided meaningful information about fracture location, 4) an improvement in scanning resolution would not improve prediction of vertebral body strength, 5) the DXA is capable to predict well femoral mechanical properties if loaded in a simulated fall and only moderately if loaded by simulating a one legged stance.
In conclusion a large dataset of experimental results (mechanical properties and 3D images) were generated and used to successfully validate QCT-based nonlinear specimen specific hvFE models of the human femur and vertebral body. The experimental results can be used in future studies to validate a number of numerical models based on QCT datasets.
Moreover, the developed hvFE models could be used without major modification for pre-clinical and clinical studies in the next future to improve the prediction of the bone strength and indirectly of the risk of fracture in vivo.