Computer tomography-based finite element analysis of the human vertebral body

Abstract

Fractures of the vertebral body, the most common type of osteoporotic fractures, seriously affect health and quality of life, and are associated with pain and increased mortality. Most vertebral fractures are asymptomatic: less than half of them are reported, and current clinical diagnosis tools and preventive methods to assess the risks of fractures lack precision. Computer tomography (CT)-based finite element (FE) methods have gained in popularity for the fields of biomechanics and orthopaedic research, with demonstrated potential to predict vertebral strength better than current risk assessment tools.<br />However, these methods still require improvements, and their accuracy is limited by methodological assumptions.<br />This thesis brings together cumulated work on improving CT-based FE methods to predict the mechanical behaviour of vertebral body. Preceded by a comprehensive introductory chapter describing the biomechanical, clinical and computational aspects related to the vertebra, this manuscript reports the studies conducted in a stepwise approach.<br />A first study pertains to the calibration of the image data used as input to the material properties for bone in the FE models. A second study addresses specific accuracy issues of FE methods to predict the apparent elastic behaviour of trabecular bone, a major constituent of the vertebral body. Anatomy-based FE models of vertebral bodies are created in a third study, to compare their mechanical behaviour with experimental tests, and to predict their structural properties and damage distributions under various loading conditions. A fourth study is performed with the same methodology to address particular questions related to a procedure called \emph{vertebroplasty}, consisting in the injection of bone cement in the vertebral body to reinforce or stabilize it. Vertebroplasty has associated risks of adjacent vertebral failures, and the presented nonlinear FE models describe the underlying mechanism behind such adverse effects, providing valuable guidelines to improve the procedure. In a fifth and last study, the FE methodology is further refined to include topologically-conform geometry and account for trabecular fabric. These models are the best to date to describe the mechanical nonlinear behaviour of vertebral bodies and provide a refined understanding of the limitations of current FE methods applied to clinical data.<br />Finally, the accomplishments of this work contribute to the improvement of current continuum finite element methods, in describing the mechanisms of damage of healthy, impaired and treated vertebral bodies, with the ultimate goal to assist in the assessment of failure risks.