Development of a subject-specific finite element model of the thoracolumbar spine segment integrating CT and MRI images.

Finite Element Analysis (FEA) is an essential tool in biomedical engineering and particularly in spinal biomechanics research. Subject-specific spine models provide a more realistic representation of load transmission, especially when there are structural changes involved, such as disc degeneration, a pathology strongly associated with chronic low back pain and mechanical instability. Under such a circumstance, it is paramount to include in the model, besides bone anatomy, the mechanical characterization of the intervertebral discs that is afforded by MRI images. The aim of this thesis is the creation of a three-dimensional patient-specific thoraco-lumbar spine model, developed from computed tomography (CT) and magnetic resonance imaging (MRI). CT images were used for segmentation and assignment of mechanical properties of vertebrae, whereas MRI was used to explore the morphology and segmentation of tissue properties of the IVDs. The integration of the two imaging modalities enabled the development of a 3D model that encompasses both anatomical accuracy and physiologic tissue response. MRI was utilized not only for morphological segmentation, but also for quantitative measurement of internal disc parameters such as water content and degeneration, which are not inferable from CT. The developed process involved semi-automatic segmentation of vertebrae in CT scans and intervertebral discs from MRI images using the 3D Slicer software. CT segmentations of the vertebras were, also, aligned to MRI reference space using an Iterative Closest Point (ICP) algorithm. The T9–T10 and T12–L1 discs, which were healthier, were separately segmented to distinguish the nucleus pulposus (NP) and annulus fibrosus (AF), whereas the T10–T11 and T11–T12 discs, which were degenerated, were modelled as homogeneous structures. For each segmented structure, T2 relaxation time was estimated using a Simulated Annealing optimization algorithm. T2 values were then used to estimate, through calibration curves, the water content and mechanical properties of the nucleus (elastic modulus and Poisson's ratio). The annulus fibrosus was modelled as a hyperelastic material using the Mooney–Rivlin formula. Where the nucleus could not be differentiated in the degenerated discs, the entire disc was regarded as AF, and the Mooney–Rivlin constants C1 and C2 were determined based on percentage disc height loss. The results indicate that T2 values correlate with the degree of degeneration: increased values in fewer degenerated discs, and lower degeneration values in degenerated discs. The elastic modulus of the nucleus increased while water content reduced, and the Poisson’s ratio reduced, indicating decreasing volumetric deformability. C1 and C2 parameters were larger in strongly height-loss discs, since it indicates more stiff mechanical behavior of the fibrous tissue. Overall, this study offers a comprehensive and reproducible workflow for spine biomechanical modelling using CT and MRI data for personalized assignment of disc material properties. The resulting model presents a solid basis for future FEM simulations regarding intervertebral disc degeneration.