Estimating lumbar passive stiffness behaviour from subject-specific finite element models and in vivo 6DOF kinematics

Passive rotational stiffness of the osseo-ligamentous spine is an important input parameter for estimating in-vivo spinal loading using musculoskeletal models. These data are typically acquired from cadaveric testing. Increasingly, they are also estimated from subject-specific imaging-based finite element (FE) models, which are typically built from CT/MR data obtained in supine position and employ pure rotation kinematics. We explored the sensitivity of FE-based lumbar passive rotational stiffness to two aspects of functional in-vivo kinematics: (a) passive strain changes from supine to upright standing position, and (b) in-vivo coupled translation-rotation kinematics. We developed subject-specific FE models of four subjects' L4L5 segments from supine CT images. Sagittally symmetric flexion was simulated in two ways: (i) pure flexion up to 12° under a 500 N follower load directly from the supine pose. (ii) First, a displacement-based approach was implemented to attain the upright pose, as measured using Dynamic Stereo X-ray (DSX) imaging. We then simulated in-vivo flexion using DSX imaging-derived kinematics. Datasets from weight-bearing motion with three different external weights [(4.5 kg), (9.1 kg), (13.6 kg)] were used. Accounting for supine-upright motion generated compressive pre-loads ≈ 468 N (±188 N) and a "pre-torque" ≈2.5 Nm (±2.2 Nm), corresponding to 25% of the reaction moment at 10° flexion (case (i)). Rotational stiffness estimates from DSX-based coupled translation-rotation kinematics were substantially higher compared to pure flexion. Reaction Moments were almost 90% and 60% higher at 5° and 10° of L4L5 flexion, respectively. Within-subject differences in rotational stiffness based on external weight were small, although between-subject variations were large.