Simulative investigation of the required level of geometrical individualization of the lumbar spines to predict fractures

Injury mechanisms of the lumbar spine under dynamic loading are dependent on spine curvature and anatomical variation. Impact simulation with finite element (FE) models can assist the reconstruction and prediction of injuries. The objective of this study was to determine which level of individualization of a baseline FE lumbar spine model is necessary to replicate experimental responses and fracture locations in a dynamic experiment. Experimental X-rays from 26 dynamic drop tower tests were used to create three configurations of a lumbar spine model (T12 to L5): baseline, with aligned vertebrae (positioned), and with aligned and morphed vertebrae (morphed). Each model was simulated with the corresponding loading and boundary conditions from dynamic lumbar spine experiments. Force, moment, and kinematic responses were compared to the experimental data. Cosine similarity was computed to assess how well simulation responses match the experimental data. The pressure distribution within the vertebrae was used to compare fracture risk and fracture location between the different models. The positioned models replicated the injured spinal level and the fracture patterns quite well, though the morphed models provided slightly more accuracy. However, for impact reconstruction or injury prediction, the authors recommend pure positioning for whole-body models, as the gain in accuracy was relatively small, while the morphing modifications of the model require considerably higher efforts. These results improve the understanding of the application of human body models to investigate lumbar injury mechanisms with FE models.