Can Multi-Vertebral CT-Based Finite Element Models Accurately Predict Strains? An In Vitro Validation Study

Many proposed FE models to predict the vertebral risk of fracture consider single vertebrae only, neglecting the role of the intervertebral discs in load transmission and distribution across vertebrae. Inclusion of the intervertebral discs in multi-vertebrae models would allow more physiological boundary conditions. However, while CT allows material properties to be assigned to the vertebrae, no information about the discs is provided. Hence, the aim of this study was to build multi-level FE models uniquely based on CT data and validate them by comparing the predicted displacements and strains against the experimental measurements. One spine segment (T10-L1) was harvested from a human spine and tested in flexion-compression in the elastic regime. During the test, displacements and strains on the anterior surface were measured with digital image correlation. The FE model was built starting from the CT scan of that same spine segment. HU-based isotropic linear elastic properties were assigned to the vertebral bone. Five different combinations of hyperelastic material properties from the literature were assigned to the discs, modeling the nucleus pulposus and the anulus fibrosus separately. The boundary conditions replicated the flexion- compression test performed experimentally. Predicted displacements and strains on the vertebrae surfaces were compared against the measured displacements and strains. The model excellently predicted the displacement field (R2 = 0.92/0.99). On the other hand, different constitutive laws for the discs resulted in different principal strain distributions, which substantially differed from the experimental one, showing average relative errors higher than 34%. In conclusion, a different modeling approach should be adopted for the discs in CT-based multi-level FE models to achieve acceptable accuracy