Preoperative planning of spinal surgery involving biomechanical analysis provides the possibility for implant customization. However, the implant customization process has been limited by excessive time or cost requirements involved in verifying implant designs using patient-specific finite element (FE) models [1]. Therefore, our project aims to build an automated pipeline to generate patient-specific FE models of the lumbar spine. A template lumbar spine FE model was established in ABAQUS 2017 based on an average-sized healthy male. A statistical shape model (SSM) was developed using a training set of 46 lumbar spines. The surface geometry of the template model was non-rigidly aligned to each spine with radial basis functions, then a principal component analysis was performed. Hence, by choosing the weightings of these principal components, the surface geometry of the template model could be fitted to a new spine [2]. Once fitted, the full FE model could be adapted to the new spine by the following steps: (a) changing the surface mesh node positions of the template FE model to match the new spine, (b) generation of solid meshes from updated surface meshes, (c) assigning material properties to generated solid meshes and (d) updating solid meshes and centre of rotations of each lumbar level in the template FE model. To evaluate the performance of FE models, lateral bending and torsion range of motion (ROM) were simulated for the template model, and for the model morphed to a representative patient (Figure-1). In each case, the simulated ROM compared reasonably to the upper and lower range to published data [3]. In conclusion, the pipeline can reduce the manual efforts of spinal surgery planning and benefit spinal implant customization. The pipeline will be further validated in other ROMs and facet joint forces, and enriched by adding spinal implant generation function.
Figure 1.