Cervical facet joints are involved in the complex movements and associated disorders of the head and neck. The capsular ligament of the cervical facet joint plays an important role in guiding and restraining joint motions under different loading modes. However, biomechanical studies on the capsular ligament to date have primarily focused on loading the ligament in tension to failure, with little understanding of the ligament response in other loading modes. Current computational human body models (HBMs) use simplified representations of the facet joint and capsular ligament. Experimental data for the multi-directional capsular ligament response could improve HBM capabilities. Therefore, the overall goal of this thesis was to investigate a hybrid experimental-computational approach to evaluate cervical facet joint mechanics under multidirectional loading and assess potential improvements to the cervical facet joint implementation in a motion segment from the GHMBC M50 neck model.

In the experimental study, ten human cervical facet joints were isolated and loaded cyclically to sub-failure displacements at 0.1 mm/s and 10 mm/s in three loading directions:​ (1) tension, (2) AP shear, and (3) LT shear. Displacement limits were determined independently for each specimen up to a maximum of 3.5 mm. Force and displacement data were recorded using a spinal loading simulator (AMTI VIVO). Additionally, a camera was used to record the displacement of a grid of markers on the capsular ligament surface. Force-displacement data for each test was separated into loading and unloading phases in the positive and negative loading directions. Two loading cycles from each specimen were extracted to calculate the average force-displacement curves using ARCGen, an arc-length re-parameterization and signal registration method. Ligament surface Green strain was estimated by digitizing the marker displacements and running them through a Matlab script. The average force-displacement curves showed the highest magnitude of force in tension, while AP shear and LT shear had similar magnitudes of force at the same displacement.

In the parallel computational study, the C4-C5 facet joint was extracted from the GHBMC M50 model. The first extraction, the FJE1 model, was used to replicate the boundary conditions used in a previous tension to failure test (Mattucci et al). The FJE1 model was also loaded in positive and negative shear. A second facet joint extraction, the FJE2 model, was setup for the multidirection experimental study described above and replicated the boundary conditions for cyclic, sub-failure tension, AP shear and LT shear at the 10 mm/s rate. The FJE2 model was then modified to examine potential enhancements to the force-displacement response. The first modification added 1D tension-only elements to the capsular ligament in a diagonal formation. The input curves for the original and diagonal elements were calculated to simultaneously fit the response in tension, positive AP and positive LT shear directions. The second modification used shell elements to represent the capsular ligament, characterized using the experimental tension data. R² values were calculated between the model output and the experimental data to monitor the improvements compared to the FJE2 model. The FJE1 model performed well in tension;​ however, the FJE2 model overestimated the force response in positive AP shear and underestimated the force response in negative AP shear and LT shear. Additionally, the 1D tension-only elements representing the capsular ligament did not interact with the hard-tissue geometry, resulting in non-physiological load paths, and the elements changed orientation before carrying load. The addition of diagonal elements improved the response in positive AP shear while maintaining good agreement in tension;​ however, the implementation was unable to improve the response in LT shear or negative AP shear. Using shell elements to represent the capsular ligament resulted in similar improvements without requiring an iterative fitting procedure and overcame some of the limitations of 1D tension-only elements.

The hybrid experimental-modelling approach facilitated the translation of new experimental data to the computational GHBMC model. The VIVO simulator allowed loading of cervical facet joints in multiple modes of loading, providing new data to evaluate the GHMBC model in AP and LT shear, which has not previously been performed. However, several key limitations should be addressed in future experimental and modelling work to further improve understanding of facet joint mechanics. These include testing the facet joint in more complex loading scenarios, investigating the effect of specimen alignment, removing the geometric bias of the 1D tensiononly elements, and improving the shell representation of the capsular ligament.