A METHOD TO GENERATE DIGITAL POPULATIONS FOR FINITE ELEMENT ANALYSIS OF SPINAL CORD INJURY FROM MRIS

Spinal cord injuries (SCIs) are a devastating outcome following a mechanical impact to the spine, which translates into the constituent tissues of the spinal cord. Since the biomechanics of human SCIs are often unknown, representative animal models are used to study the injury, and computational models are used to study the injury biomechanics. Morphological variability is a key influencing factor in SCI outcomes and is inherently captured in both human SCI incidents and animal experiments, but computational models of SCI often investigate findings in a single geometry and thus, morphological variability is neglected. In this study, we present an approach to incorporate morphological differences into a computational model of a unilateral contusion injury in nonhuman primates (NHP). Pre-injury MRIs of subjects [Formula: see text] were parameterized using four parameters, including the diameters of the cord and canal in both the mediolateral and anterior-posterior axes and inputted into a Python code to generate new geometries within the measured range. The code was used to generate a digital population [Formula: see text] and their average morphology and variability matched that seen in the real NHP subjects. A subset of the geometries [Formula: see text] was used to simulate a unilateral contusion, and the predicted peak forces [Formula: see text] were close to values seen in NHP experiments. Morphological variability alone accounted for a 10% variability in the peak forces experienced by the cord. This study highlights the importance of using multiple geometries in computational models to ensure findings are more robust and clinically translatable.