A Novel 3D/2D Correspondence Building Method for Anatomy-Based Registration

摘要

The application of fluoroscopic images in operation is pervasive, especially for orthopaedic surgery. Anatomy-based 3D/2D registration, rigid or non-rigid, has been proven to improve the accuracy and precision of various image-guided therapies. One of the key steps for a successful anatomy-based registration is to establish 3D/2D correspondence between the 3D model and the 2D images. This paper presents a novel 3D/2D correspondence building method based on a non-rigid 2D point matching process, which iteratively uses a symmetric injective nearest-neighbor mapping operator and 2D thin-plate spline based deformation to find a fraction of best matched 2D point pairs between features detected from the X-ray images and those extracted from the 3D model. The estimated point pairs are further ranked by their shape context matching cost and those with high cost are eliminated. The remaining point pairs are then used to set up a set of 3D point pairs such that we turn a 3D/2D registration problem to a 3D/3D one, whose solutions are well studied. Rigid and non-rigid registration algorithms incorporating the novel 3D/2D correspondence building method are presented. Quantitative and qualitative evaluation results are given, which demonstrate the validity of our method.