Abstract
Introduction
Although weight-bearing CT of the foot definitely reflects the morphology and deformity of joint, it is hard to obtain the standing CT due to difficulty of availability. Although 3D imaging reconstruction using radiographs has been reported in other joints, there is no study about foot joint. The purpose of this study is to develop a semi-automatic method based on a deformable surface fitting for achieving the weight-bearing 3D model reconstruction from standing radiographs for foot.
Methods
Our method is based on a Laplacian surface deformation framework using a template model of foot. As pre- processing step, we obtained template surface meshes having the average shapes of foot bones (talus, calcaneus) from standing CT images (Planmed Verity) in 10 normal volunteers. In the reconstruction step, the surface meshes are deformed following guided user inputs with geometric constraints to recover the target shapes of 30 patients while preserving average bone shape and smoothness. Finally, we compared reconstructed 3D model to original standing CT images. Analysis was performed using Dice coefficients, average shape distance, maximal shape distance.
Results
The obtained reconstruction model is close to the actual standing foot geometry (Dice coefficients 0.89, average shape distance 0.88 mm, maximum shape distance 6.33 mm). We present the accuracy and robustness of our method via comparison between the reconstructed 3D models and the original bone surfaces.
Conclusions
Weight-bearing 3D foot model reconstruction from standing radiographs is concise and the effective method for analysis of foot joint alignment and deformity.
