AutoBend: An Automated Approach for Estimating Intervertebral Joint Function from Bone-Only Digital Models

Deciphering the biological function of rare or extinct species is key to understanding evolutionary patterns across the tree of life. While soft tissues are vital determinants of joint function, they are rarely available for study. Therefore, extracting functional signals from skeletons, which are more widely available via museum collections, has become a priority for the field of comparative biomechanics. While most work has focused on the limb skeleton, the axial skeleton plays a critical role in body support, respiration, and locomotion, and is therefore of central importance for understanding broad-scale functional evolution. Here, we describe and experimentally validate AutoBend, an automated approach to estimating intervertebral joint function from bony vertebral columns. AutoBend calculates osteological range of motion (oROM) by automatically manipulating digitally articulated vertebrae while incorporating multiple constraints on motion, including both bony intersection and the role of soft tissues by restricting excessive strain in both centrum and zygapophyseal articulations. Using AutoBend and biomechanical data from cadaveric experiments on cats and tegus, we validate important modeling parameters required for oROM estimation, including the degree of zygapophyseal disarticulation, and the location of the center of rotation. Based on our validation, we apply a model with the center of rotation located within the vertebral disk, no joint translation, around 50% strain permitted in both zygapophyses and disks, and a small amount of vertebral intersection permitted. Our approach successfully reconstructs magnitudes and craniocaudal patterns of motion obtained from ex vivo experiments, supporting its potential utility. It also performs better than more typical methods that rely solely on bony intersection, emphasizing the importance of accounting for soft tissues. We estimated the sensitivity of the analyses to vertebral model construction by varying joint spacing, degree of overlap, and the impact of landmark placement. The effect of these factors was small relative to biological variation craniocaudally and between bending directions. We also present a new approach for estimating joint stiffness directly from oROM and morphometric measurements that can successfully reconstruct the craniocaudal patterns, but not magnitudes, derived from experimental data. Together, this work represents a significant step forward for understanding vertebral function in difficult-to-study (e.g., rare or extinct) species, paving the way for a broader understanding of patterns of functional evolution in the axial skeleton.