Integrating muscle forces from neuro-musculoskeletal modeling into an osseo-ligamentous finite element model of the spine

INTRODUCTION The computational methods of finite element (FE) analysis and neuro-musculoskeletal (NMS) modelling have both historically been used as an effective tool in simulating and improving understanding of the biomechanics and dynamic motion of the human spine. Studies integrating muscle-driven boundary conditions into FE simulations of the human spine are, however, lacking. The aim of this project is to integrate these two synergistic methods to garner a greater understanding of the impact of spinal muscles on mechanics in the thoracolumbar spine during forward flexion. METHODS Using custom image processing software, Dicomtilt, FE and NMS models of a thoracolumbar spine with pelvis were individualized using 3D CT images of a male osseous spine (NIH Visible Human Project). 'Demoa', an established custom-coded NMS modelling algorithm, was used to extract the maximum contractile force for 132 muscle fascicles acting across the thoracolumbar spine during forward flexion. These muscles included major muscle groups – erector spinae, rectus abdominus, external/internal obliques. Individual muscle fascicles were modelled as axial connector elements and the NMS-derived muscle forces allocated to the muscle fascicles as a connector load, linearly ramped over the load step. To determine the difference in predicted spine mechanics between a muscle-activated FE loading condition and traditional approaches to quasi-static, displacement/force controlled spinal FE loading, three additional loading conditions were simulated, representing similar forward flexion motion. These were as follows: (1) A displacement controlled forward flexion of 7 degrees applied about an instantaneous axis of rotation, anterior to the T7 vertebra; (2) An anteriorly located body weight force applied 50mm from the T1 vertebra; (3) An anteriorly located bending moment as per (2), with an additional compressive follower load acting throughout the thoracolumbar spine, to represent the stabilising contribution of the spinal muscles. FEM (1) and (2) included the 132 spinal muscle fascicles, with a passive stiffness only. All simulations were initiated from a comparable state, using intrinsic hydrostatic pressure in the intervertebral disc. The model was spatially constrained by fixing the sacrum in all degrees of motion. RESULTS AND DISCUSSION Results demonstrated a difference in the tissue-level mechanics with the four different loading methods, despite the overall spine exhibiting similar degrees of forward flexion. These results imply the need for a thorough consideration of the type and location of boundary conditions used to computationally represent physiological motion. CONCLUSION Future studies will include additional physiologically relevant structures from NMS modelling, to investigate the impact of individual muscle groups on the movement of the thoracolumbar spine. Application for patient specific simulations to predict the effects of atrophy or trauma to different muscle groups of the spine will be possible.