A study of impact loading of the spine

Abstract

Cervical Spine Injuries (CSIs) arising from collisions are rare in contact sports, such as rugby union, but their consequences can be devastating as they can lead to paraplegia, tetraplegia and death. Finite Element (FE) models have been widely used to give a more in-depth understanding of the biomechanics of injury. Several FE modelling approaches are available in the literature, but a fully calibrated and validated FE modelling framework for cervical spines under compressive dynamic loading is still lacking and material properties are properly not calibrated for such events. This study developed and validated a methodology for specimen-specific FE modelling of the cervical spine under impact loading.
Thirty-five (n=35) individual vertebral bodies (VBs) and three (n=3) whole cervical spines (C2 to C7) were dissected from porcine spine segments. Samples were potted in bone cement, mCT scanned and a speckle pattern was applied to the anterior aspect of the bones to allow Digital Image Correlation (DIC). Surface displacements and strains were acquired using DIC. Twenty-seven (n=27) VBs were compressively tested to a load up to 10kN applied at a rate of 1000N/min from the cranial side. Specimen-specific FE models were developed for fourteen (n=14) samples in this group and the material properties were optimised based on the experimental load-displacement data and using a factor (KGSStatic ) to calibrate a density to Young’s modulus equation. Thirteen (n=13) FE models were created from the remaining tested samples: the previously optimised density to Young’s modulus relationship was applied to this group and the resulting vertebral stiffness was compared to experimental findings. This allowed validation of the developed density to Young’s modulus relationship. Eight (n=8) remaining VBs were subjected to an impact load applied via a falling mass of 7.4 kg at a velocity of 3.1ms􀀀1. Surface displacements and strains were acquired from the anterior VB surface via DIC, and the impact load was monitored with two load cells. Specimen-specific FE models were created for this group and material properties were assigned again based on the density-Young’s modulus relationship previously validated for static experiments, supplemented by an additional factor (KGSDynamic ) derived from an iterative comparison between numerical stiffness predictions and experimental findings. Three (n=3) whole cervical spines were subjected to the spine impact loading conditions and surface displacements, strains and force profiles were also acquired. Spine specimen-specific FE models were created for these samples using the previously developed modelling workflow. VBs material properties were assigned via KGSDynamic , grey-scale and the density-Young‘s modulus relationship. Intervertebral discs were defined as homogeneous and isotropic and mechanical properties were assigned to the specimens using a kinematic approach. Experimental and numerical load-displacement curves for these spinal segments were compared using Bland-Altman plots, non-parametric tests and Lin’s concordance coefficient (CCC). The optimised conversion factor for quasi-static loading, KGSStatic , had an average of 0.033. Using this factor, the validation models presented average numerical stiffness value 3.72% greater than experimental. From the dynamic loading experiments, the value for KGSDynamic was found to be 0.14, 4.2 times greater than KGSStatic . Average numerical stiffness was 2.3% greater than experimental. Almost all models presented similar stiffness variation and regions of maximum displacement and strains similar to what observed via DIC. Spine models presented similar biomechanical behaviour to experimental data in terms of predicted peak load, vertebral displacements and strains and three-dimensional spine movements. Strain patterns exhibited a high level of similarity to those recorded via DIC but strain magnitudes were consistently smaller. Deformation by buckling was observed. The developed FE modelling methodology allowed the creation of models which predicted both static and dynamic behaviour of VBs. The same methodology was then applied to the modelling of the whole cervical spine, which resulted in predictions showing agreement with experimental data. Strains and deformation patterns were assessed and evaluated, showing that buckling was the main deformation mode for this type of impact. This methodology is now validated to be fully applied to simulate axial impact scenarios replicating rugby collisions events.

Date of Award	20 Jan 2020
Original language	English
Awarding Institution	University of Bath
Supervisor	Sabina Gheduzzi (Supervisor) & Richie Gill (Supervisor)