A Hybrid Modeling Workflow for Performing Finite Element Analyses Under In-Vivo Conditions

Abstract

Computational models are commonly used to evaluate spinal loads due to difficulties associated with acquiring in-vivo measurements. Musculoskeletal (MSK) models can be used for performing simulations of movement by considering active structures such as muscles and simplified representations of joint structures. Finite element (FE) models are rather actuated with simplified loading conditions determined from in vitro studies and consider detailed representations of tissues and joint structures, making them suitable to evaluate stresses and strains within the spine. This work was done in collaboration with Numalogics Inc., where a hybrid modeling workflow was developed to combine MSK and FE modeling methods to be able to evaluate stresses and strains within spinal instrumentation devices under the influence of in-vivo loading conditions, i.e., muscle forces. A previously validated fully articulated thoracolumbar spine MSK model was selected, simplified, and replicated exactly in Ansys, the FE modeling platform. The sensitivity of the compressive joint loads to the modifications brought to the original model was evaluated for static flexion poses. Then, two hybrid simulation workflows were developed. Version I consisted of performing a static optimization analysis using the MSK model to calculate muscle forces that satisfy equilibrium constraints. Version II consisted of performing a stability-constrained static optimization analysis to calculate muscle forces and stiffnesses that satisfy equilibrium and stability constraints in the MSK model. The forces and moments calculated from the MSK model were then transferred to the FE model as loading conditions. Validation studies were performed for both simulation workflows by comparing the kinematics and compressive joint loads in the MSK and FE models under identical loading conditions, for various static flexion poses and for six scaled models. Version I could not be validated due to convergence issues and large discrepancies between model outputs. Version II was successfully validated, with negligible kinematic and joint load differences between models across different ranges of motion. Very strong to perfect correlations were observed between model kinematics and joint loads, respectively, over all simulations performed. Overall, it was found that the consideration of stability constraints is essential for successfully performing FE analyses of the entire thoracolumbar spine under in-vivo loading conditions over physiological ranges of motion. In this work, no soft tissue deformations nor stresses were evaluated with the FE model. However, with the developed workflow, Numalogics will be equipped with a framework for their future works, including the comparison of stress distributions in spinal instrumentation devices when applying in vitro versus in-vivo loading conditions to their FE models.