CCTA 영상에서 계층적 변형을 이용한 심혈관 비강체 정합

초록

In this paper, we propose a accurate and fast non-rigid registration method between vessels using two 3D cardiac computed tomography angiography(CTA) images taken at different times to provide auxiliary information that can be used in the diagnosis and treatment of coronary artery disease. The proposed method consists of the following four steps. First, global alignment is performed through rigid registration between 3D vessel centerlines acquired from two 3D CTA images. Second, corresponding points is obtained through point matching between 3D vessel centerlines. Third, the outlier is removed from the corresponding points using the thickness and gradient of the vessel. Fourth, non-rigid deformation for hierarchical local transformation is performed using an energy function. As a result of the experiment, the average registration error is 0.987 mm, and the average execution time is 2.137 seconds. The proposed method shows that the registration is performed accurately and quickly.

키워드