Novel Looped-Catheter-Based 2D-3D Registration Algorithm for MR, 3DRx and X-Ray Images: Validation Study in an Ex-vivo Heart

详细信息    查看全文

• 关键词：2D ; 3D registration ; Cardiac image registration ; Image ; guided procedures ; MR ; 3DRx ; X ; ray fluoroscopy ; Biplane X ; ray
• 刊名：Lecture Notes in Computer Science
• 出版年：2017
• 出版时间：2017
• 年：2017
• 卷：10124
• 期：1
• 页码：152-162
• 丛书名：Statistical Atlases and Computational Models of the Heart. Imaging and Modelling Challenges
• ISBN：978-3-319-52718-5
• 卷排序：10124

文摘

In this paper, a novel 2D-3D cardiac image registration algorithm is proposed for application in X-ray-guided catheterisation procedures, and relies on a common technique of inserting a catheter and then looping it inside a chamber of the heart for visual reference. Registration starts with the isocentre-supine constraint and then iteratively refined by maximising a feature-based area metric using an inserted catheter loop and the segmented cardiac border from one or more X-ray views. Maximisation is done in two stages: first correcting for translational motion, and then simultaneously correcting for rotations and translations. The two-staged approach is demonstrated to be more accurate than a similar single-staged approach in an explanted porcine heart. In this experiment, accuracy was demonstrated to be within the 5-mm clinical requirement. On average, the algorithm could register images with a mean target registration error (TRE) of 4.6-mm when using two X-rays (biplane), and a mean reprojection distance (RPD) of 1.9 mm using a single view (monoplane).