Ultrasound Navigation for Transcatheter Aortic Stent Deployment Using Global and Local Information

Abstract

An ultrasound (US) navigation system using global and local information is presented for transcatheter aortic stent deployment. The system avoids the use of contrast agents and radiation required in traditional fluoroscopically-guided procedures and helps surgeons precisely visualize the surgical site. To obtain a global 3D (three-dimensional) navigation map, we use magnetic resonance (MR) to provide a 3D context to enhance 2D (two-dimensional) US images through image registration. The US images are further processed to obtain the trajectory of interventional catheter. A high-resolution aortic model is constructed by using trajectory and segmented intravascular ultrasound (IVUS) images. The constructed model reflects morphological characteristics of the aorta to provide local navigation information. Our navigation system was validated using in vitro phantom of heart and aorta. The mean target registration error is 2.70 mm and the average tracking error of the multi-feature particle filter is 0.87 mm. These results confirm that key parts of our navigation system are effective. In the catheter intervention experiment, the vessel reconstruction error of local navigation is reduced by 80% compared to global navigation. Moreover, the targeting error of the navigation combining global and local information is reduced compared to global navigation alone (1.72 mm versus 2.87 mm). Thus, the US navigation system which integrates the large view of global navigation and high accuracy of local navigation can facilitate transcatheter stent deployment.