Clinical-oriented spinal surgical planning based on finite element method and automate-generated surgical templates assisting the spinal surgery

Abstract

Osteoporosis and inappropriate implant configurations increase the risk of mechanical failure in bone-implant fixation system, such as screw loosening, fractures of spinal rods and refracture after vertebral augmentation. Currently, biomechanical evaluation of bone-implant system was widely used to predict the stress distribution, underlying implant failure and long-term fatigue performance, and finite element analysis (FEA) was the most common approach. However, reported research work were mechanical-oriented, in which non-linear material models of soft tissues were implemented to capture stress pattern in intervertebral discs (IVDs); thus, the non-linearity required high computational cost and lead to diverging problem, even analysis failure with high possibility. This study aimed to develop a clinical-oriented finite element (FE) modelling strategy for spinal fixation evaluation, which took the patient-specific anatomy into consideration. Then an automatic workflow of surgical template generation was proposed together with adapted computer graphical algorithms. In terms of FE modelling, a phantom-less bone mineral density (BMD) approach was employed to measure the material property of spinal units, thereby individual bony features were considered. Spinal fixation system was mounted to restrict the range of motion (ROM) of vertebra segments, resulting in less load sharing in IVDs. Thus, biomechanical response in IVDs were not the major concern in evaluation of bone-implant system, and simplification was allowed to lowdown the non-linearity of the FE model. Additionally, a practical damage indicator of bone was defined as a ratio of principal strain, considering the deformation of resultant forces. The indicator was validated against a retrospective case of ACDF (anterior cervical discectomy fusion) and showed good consistency. A following application of the proposed FE model was to compare an innovative suspensory traction with traditional Halo traction in patients with cervical kyphosis, demonstrating a better relaxation of anterior soft tissues was achieved in suspensory traction and good protection on the spinal canal from over-extension. Furthermore, the proposed FE modelling strategy was implemented to evaluate the biomechanical response of L2 – L4 with six pedicle screws and two intervertebral spacers. The FE results gave a possible reason of screw loosening - lateral force (perpendicular to screw axis) on screws pushed screws towards bony structures in various directions randomly, causing bone within threads subjected to shear stress repeatedly. Similar load sharing was observed in ACDF as well, suggesting that the biomechanical evaluation should focus on the entire bone-implant system and fully consider the coupling effect among screws introduced by spinal rods. As for the generation of patient-specific surgical templates, fundamental algorithms of border smooth, offset, generation of standard geometry and Boolean operation were developed. Geodesic curvature was imported as optimization objective, while smooth coefficient and deviation allowance were defined to control the iteration loop. The algorithm smoothed the input curve without unacceptable deviation. The offset algorithm was developed with normal vector of vertices and iterative bisection, outputting a solid layer of elements based on input triangle mesh and was validated against non-linear surface in vertebra body. Remaining algorithm needed was reimplemented from open-source libraries.