Computational models for the pre-operative planning of spinal surgeries

Spinal surgeries are common to treat different types of diseases and injuries as trauma, tumours, deformity and degenerative diseases. Conventional open spine surgery has several reported limitations and there is a trend towards minimally invasive techniques due to lower complication rates and morbidity. Percutaneous pedicle screw fixation and vertebral augmentation are two widespread minimally invasive techniques that are often chosen to treat vertebral fractures. The pre-planning of spinal surgeries is based on anatomical measurements taken on clinical images and on the experience of surgeons. Post-operative complications may arise impacting the quality of life of patients. Computational models can provide important patient-specific information about the biomechanics and the geometry of the spine.
Finite element (FE) models have the potential to predict the biomechanical outcomes of surgeries and are often proposed as possible tools for planning pedicle screw fixation. However, before their application, these models have to be verified and the sensitivity of metrics used to assess metal and bone failure have to be assessed with respect to the screw size and geometry. This was the aim of the first study. Patient-specific Computed Tomography (CT)-based FE models of the human vertebra with two pedicle screws were verified for both realistic and simplified geometry of screws. The diameter of the screw played a major role on the mechanics of the screw-vertebral structure with respect to the length. Simplified screws could accurately estimate the deflection and the strain of the implanted vertebrae, but resulted in a systematic underestimation of the peak stress in the screws. FE models can be used to optimize surgery-related parameters, but take a long time to compute and are thus insufficient to fulfil the demands of most clinical settings. Reduced Order Models (ROMs) are useful tools to improve the efficiency of FE models, and, in this thesis, were applied to FE models of the implanted vertebra to optimise screws’ size and orientation showing accurate prediction of the deflection and the stress of the screws. A third study included the development of a CT-scan based procedure to estimate the pre-fracture 3D shape of a L1 vertebra that could be used by surgeons to restore the pre-fracture biomechanics. The methodology was validated on a dataset with 40 patients and showed excellent reconstruction accuracy.
In conclusion, FE models of the implanted vertebra were integrated with ROMs to build a computational pipeline for the optimisation of dimensions and positioning of pedicle screws. Also, the geometric pre-fracture shape of a L1 vertebra was reconstructed. These approaches can be used to provide more quantitative biomechanical and geometric information to surgeons for planning the treatment of vertebral fractures.