Mechanical Characterisation and Computational Modelling of Spinal Ligaments

Low back pain is a common complaint in people of all ages. The long-term success rates of many surgical devices to treat the spine have been relatively low and improved methods of pre-clinical testing of these devices are therefore needed. Sheep spine models are commonly employed in pre-clinical research studies for the evaluation of spinal devices. The anterior and posterior longitudinal ligaments (ALL and PLL) provide passive stability to the spine, however, limited studies have been conducted to characterise the mechanical properties of the ovine longitudinal ligaments or compare them to the human. Moreover, previous studies have derived material properties for the human ALL and PLL directly from force-displacement data, assuming uniform cross sectional area and length, and these values have been used extensively in finite element models of the spine for the analysis of clinical interventions. The aim of this study was to develop a methodology to test and compare the stiffness of human and ovine spinal longitudinal ligaments and to uniquely combine experimental and specimen-specific finite element (FE) modelling approaches to determine the ligament mechanical properties. The methodology was developed on ovine thoracic spines and then applied to human thoracic spines. The spines were dissected into functional spinal units (FSUs) with the posterior elements removed and imaged under micro computed tomography (µCT). The specimens were sectioned through the disc to leave only either the ALL or PLL intact and tested in tension to determine the stiffness. The µCT images from each FSU were used to build specimen-specific FE models of the ligaments and bony attachments. Hyper-elastic material models were used to represent the ligament behaviour. Initial values for the material model were derived using mean cross sectional area (CSA) and length (L), with the assumption that ligament was uniaxially loaded. The parameters were then iteratively changed until a best fit to the corresponding experimental load-displacement data was found for each specimen. The stiffness of the ligaments for the ovine specimens were found to be higher than for the human specimens. This may have implications for the use of ovine FSUs for preclinical testing of devices. There was poor agreement between the material parameters derived from FE models and the initial values derived by assuming a mean CSA and L. This work demonstrates that a specimen-specific image-based approach needs to be applied to derive the elastic properties of the ligaments due to their non-uniform shape and cross-sectional area.