Search for a command to run...
One third of cancer patients present vertebral metastases with possible neurological complications. SINS, a radiographic scoring system based on clinicians' visual estimates, is the current standard to evaluate the instability of a metastatic spine, despite its qualitative nature. This thesis is the starting point of project that aims to introduce in the SINS a personalized assessment of vertebral biomechanics, by means of subject-specific finite elements models (SSFE) built from CT images.The heterogeneity of modeling approaches and results present in the literature, indicates that no consensus has yet been reached, neither in intact nor in metastatic vertebrae. Starting from this consideration, a novel validation experiment was designed in the project. The objective of this thesis was to explore different structural and material modeling variables within a SSFE modeling procedure developed to estimate the vertebral stiffness, strains, and strength measured in the mechanical experiment. To this aim: -a CT-based SSFE modeling procedure was defined synergistically with the experimental set-up.Some choices of the experimental set-up were just replicated in the FE model, while some other choices were driven by the modeling procedure developed in this thesis. -a non-linear SSFE modeling procedure, chosen from the literature as gold standard for its excellent validation results, was replicated (Imai et al, 2006). The features of this SSFE are a shell that mimics the cortical layer, and a pressure sensitive Drucker-Prager yield criterion coupled to bilinear plasticity. Both required significant effort to be replicated within the FE software available in the laboratory (Ansys). -modeling variables which are still inconsistent in the literature and may critically affect the results were identified: (i)material modeling (ii)use of an a-priori defined shell to mimic cortical bone. In our exploratory SSFE design we considered the combination of shell presence/absence and two density-elasticity relationships (Imai et al., 2006 and Morgan et al. 2003). We tested: the effect of a filter to avoid double-counting of cortical-like densities in shell-bearing models; and the effect of adding bone transverse isotropy. All models were realized from CT scan of a human L4 sample, which will be the first to undergo mechanical testing. The comparison of results from the models built highlighted: -a very small contribution of the Drucker-Prager yield/plasticity model to the overall failure simulation, so that estimated failure loads from the reference model and those from the linear model were fully comparable. Usage of a linear model is therefore suggested until falsified by experimental validation; -a small contribution of the cortical bone modeling through a thin shell, unless inner material models are filtered to exclude double counting of cortical-like densities. This latter implementation is thus strongly advised, and a call is launched for subject specific estimates of cortical bone properties; -a possible overestimate of stiffness and strength by the reference model and the femur-wise linear model. This is a hypothesis based on ranges of failure loads taken from literature. In fact, only the softest modeling combinations attempted returned stiffness and strength values comparable to those published. In conclusion, this thesis has defined a SSFE modeling framework including variations to be tested in a novel validation experiment on human vertebrae.