CT based nite element models of proximal femur: application to clinical studies and in vitro validation with aid of images post processing algorithms

Fractures are the clinical endpoint of osteoporosis. However, the current standard of care used clinically (areal bone mineral density (aBMD) measured by Dual X–ray Absorptiometry) is limited with respect to identifying cases at high risk of fracture, as it does not take into account many mechanical determinants that concur to bone fracture. Improvements in the assessment of bone fracture risk is an important research area with clear clinical relevance. Patient–specic numerical models of bone segments from Computed Tomography images are widely used in biomechanics research. These models can better model the personalised mechanical determinants of fracture, and may be able to overcome aBMD limitations in identifying cases at high risk of fracture. One goal of this approach is to move these models to clinical practice. However, prior to clinical translation, two requirements are necessary: i) in–vitro validation of numerical models predictions, and ii) assessment of their performance in clinical studies. Acceptable in–vitro validation of nite element (FE) models of the proximal femur for the prediction of strains and failure loads has been reached by several studies. The rst clinical applications of such models have recently been reported in case–control studies of bone osteoporotic fractures. In particular, these studies have been focused on the association of CT–based FE estimates of femoral strength with fracture status compared to the association with fracture status of the standard of care used in the clinical practice (aBMD). The results among all studies were promising, albeit heterogeneous. In the past few years, a linear, fully dened, subject–specic FE modeling procedure was developed in the Medical Technology Laboratory (Rizzoli Orthopaedic Institute, Bologna, Italy) which has been validated in–vitro for bone strains, failure load and failure location. Overall, this method yielded very good results in comparison to in–vitro experimental measurements, although some discrepancies between FE results and experimental values still exist. The rst aim of this thesis was to verify the ability of this FE modelling procedure in classifying osteporotic fractures compared to the ability of aBMD in three case–control clinical studies: a cross–sectional and a prospective study on proximal femur fractures and a cross–sectional study on prevalent osteporotic fractures of any type. The results have shown that the femur FE strength estimates can add complementary value to the aBMD in elderly osteopenic/osteoporotic women, the population at highest risk. The results available in men suggested a gender difference in FE predictive ability that requires further investigation in a larger cohort. Site–specicity was found to be important: the femur models predicted femur fractures, but not prevalent fractures at any skeletal site. Finally, the good performance of minimum strength derived from multiple loading conditions highlighted that a wider consideration of loading conditions can improve FE model performance. The second aim of this thesis was to explore possible improvements in the FE methodology based on the correction of partial volume effect (PVE) in the CT data, which can lead to errors in the bone surface geometry and intensity. This was addressed by including a CT image post–processing algorithm designed to restore geometry and intensity of thin bone structures into FE workow. Results showed that the incorporation of PVE reduction methods can lead to improvements in the FE predictions of strains and failure loads as compared to in–vitro experimental data. The CT–based FE technique, which combines biomechanical principles and patient–specic information derived from clinical CT data, can provide an alternative clinical tool to identify cases at high risk of fracture. Improvements on FE methodology focused on better capturing all mechanical determinants that concur to bone fracture may further improve assessment of fracture risk.