A pipeline to morph finite element models of the lumbar spine for generation of customised spinal cages.

Abstract

Background: Lumbar interbody fusion with spinal cages is an established treatment for degenerative disc disease, however, generic cages comprising standardised shapes and sizes still encounter complications such as cage migration and subsidence. Incorporating biomechanical analysis and implant customisation for pre-operative planning provides the possibility to optimise cage positioning and its 3D geometry, such that cage loading is optimised to reduce the chance of post-operative complications. However, current customisation processes involve considerable resources to develop personalised finite element models for evaluating bespoke cage designs. Therefore, this study aimed to build an automated pipeline to generate personalised lumbar FE model with a customised spinal cage.

Methods: A template lumbar FE model was established based on an average-sized healthy male, whose surface mesh was non-rigidly aligned to a training set of 46 lumbar spines. Then, a statistical shape model was built to predict the shape of new spines by changing the weightings of parameters inside, which facilitate the generation of personalised FE model. Once the spine mesh was fitted to the new subject, a customised cage was created by extruding a 2D cage template in the axial direction then trimming it via a Boolean function with the adjacent endplates. After adjustment of cage parameters, the cage model could be added to the FE model to replace one disc component. The range of motions and facet joint forces in typical lumbar rotations were extracted from the results of FE analyses.

Results: The predicted ROM and facet joint force for corresponding intact lumbar models compared well with or approached to the range of published data, while the cage insertion significantly reduced the ROM of lumbar spine in all directions (all p[Formula: see text]0.02). Also, the lateral cage position affected the ROM of the whole spine and resulted in larger fracture volume on the vertebrae based on the analysis on a specific subject.

Conclusions: The pipeline can generate personalised lumbar FE models and customised spinal cages with predictive outputs. Its automatic process reduced the effort expenditure involved in FE model development and cage design, making it viable for application in fusion surgery planning and spinal cage customisation.