Methodology for the assessment of the risk of failure of metastatic vertebrae through ROM-based patient-specific simulations

The structural performance of a vertebra can be significantly undermined if it develops a tumour, that could even lead to the vertebra's structural collapse. In cancers with a high prevalence of spinal metastasis, like prostate and breast cancer, this supposes an additional problem to account for on top of the regular treatment. In this work, we propose a patient-specific methodology capable of immediately predicting the structural behaviour and risk of failure of a vertebra with a spherical tumour of arbitrary characteristics. This immediate evaluation of the results, together with the ease of use of the proposed methodology, that includes the creation of a personalized computational model from a CT scan of the patient's vertebra, makes this methodology suitable for its use in clinical practice. By running several personalized structural analyses of vertebrae with simulated tumours, using the Cartesian grid FEM (cgFEM) in combination with the Sparse Subspace Learning (SSL) technique, we generate a surrogate model of the vertebra. This model is able to predict the vertebra's behaviour for different tumour growth scenarios, and could be useful as a clinical decision support tool.