Lower cervical spine facet cartilage thickness mapping

Finite element (FE) models of the cervical spine have been used with increasing geometric fidelity to predict load transfer and range of motion (ROM) for normal, injured, and treated spines. However, FE modelers frequently treat the facet cartilage as a simple slab of constant thickness, impeding the accuracy of FE analyzes of spine kinematics and kinetics. Accurate prediction of facet joint contact forces and stresses, ROM, load transfer, and the effects of facet arthrosis require accurate representation of the geometry of the articular cartilage of the posterior facets. Previous research has described the orientations of the facet surfaces, their size and aspect ratio, and mean and maximum thickness. However, the perimeter shape of the cartilaginous region and the three-dimensional distribution of cartilage thickness remain ill-defined. As such, it was the intent of this research to further quantify these parameters.

Vertebrae from seven fresh-frozen unembalmed human cadavers were serially sectioned and the osteochondral interface and the articulating surface of each facet on each slice were identified. The cartilage thickness was recorded at nine equidistant points along the length of each facet. It was observed that facets tended to have elliptic or ovoid shapes, and best-fit ovoid perimeter shapes were calculated for each facet. The thickness distribution data were used to represent the entire three-dimensional cartilage distribution as a function of one variable, and a thickness distribution function was optimized to fit the thickness distribution. The antero-posterior and medial/lateral shifts of the thickness center relative to the geometric were calculated and reported.

High correlation was observed between the ovoid perimeter shapes and the measured facet shapes in radial coordinates, indicating that the ovoid approximation is able to accurately represent the range of facet geometries observed. High correlation between the measured and fitted thickness distributions indicates that the fitting function used is able to accurately represent the range of cartilage thickness distributions observed.

Utilization of a more physiologic cartilage thickness distribution in FE models will result in improved representation of cervical spine kinematics and increased predictive power. The consistency observed in the thickness distribution function in this study indicates that such a representation can be generated relatively easily.