Use of a Dense Surface Point Distribution Model in a Three-Stage Anatomical Shape Reconstruction from Sparse Information for Computer Assisted Orthopaedic Surgery: A Preliminary Study

摘要

Constructing anatomical shape from extremely sparse information is a challenging task. A priori information is often required to handle this otherwise ill-posed problem. In the present paper, we try to solve the problem in an accurate and robust way. At the heart of our approach lies the combination of a three-stage anatomical shape reconstruction technique and a dense surface point distribution model (DS-PDM). The DS-PDM is constructed from an already-aligned sparse training shape set using Loop subdivision. Its application facilitates the setup of point correspondences for all three stages of surface reconstruction due to its dense description. The proposed approach is especially useful for accurate and stable surface reconstruction from sparse information when only a small number of a priori training shapes are available. It adapts gradually to use more information derived from the a priori model when larger number of training data are available. The proposed approach has been successfully validated in a preliminary study on anatomical shape reconstruction of two femoral heads using only dozens of sparse points, yielding promising results.