Development of an Advanced 4D Foot Shape Model from Multiple Acquisitions at Different Flexion Angles

Abstract

A major challenge in constructing models of the human body or body parts stems from the fact that they are actually composed of several solid bodies articulated together. Consequently, 3D models must be parameterised with reference to joint angles: scans of the foot taken at different plantar/dorsal flexion angles were used in this study as a benchmark. The implemented methodology was based on the regression fitting of landmark trajectories obtained on a limited number of foot poses; these same landmarks were then used to guide 3D mesh morphing to predict foot geometry at different plantar/dorsal flexion angles. A careful optimisation of the methodology was performed to identify the optimal set of foot scans. The comparison between the actual foot shapes obtained by 3D laser scanning and the predicted shapes showed that the average error is at most 6.57 ± 2.74 mm. The methodology indications were finally drawn based on these error estimates. Further works will consider combined motions including flexion/extension, internal/external rotation, and foot inversion/eversion.