3D CT to 2D X-Ray Image Registration for Improved Visualization of Infrapopliteal Vessels in Endovascular Procedures

Abstract

Chronic limb-threatening ischemia resulting from advanced peripheral arterial disease in infrapopliteal arteries poses a challenge for revascularization, with endovascular interventions suffering from a high technical failure rate. Image fusion technology can enhance the X-Ray Fluoroscopy (XRF) imaging used to guide these procedures, by incorporating vascular maps from pre-interventional Computed Tomography (CT) angiography. This work presents an image registration method designed to facilitate fusion imaging by performing automatic alignment using a single XRF view, based on intrinsic bone landmarks. The method uses an iterative optimization approach, a custom similarity function to quantify alignment, and various strategies to improve capture range and computation time. The proposed method is evaluated on a dataset of images approximating those expected in practice, consisting of cone-beam CT and XRF images of ex-vivo limbs. The results demonstrate sufficient accuracy to identify guidewire deviations from the path of the artery, with average misalignments of <1 mm at the artery.