Accurate Extraction of Coronary Vascular Structures in 2D X-ray Angiogram Using Vascular Topology Information in 3D Computed Tomography Angiography

Abstract

Since the 2D X-ray angiogram enables the detection of vascular stenosis in real-time, it is essential for percutaneous coronary intervention (PCI). However, the accurate vascular structure is very difficult to determine due to background clutter and loss of depth information of 2D projection. To cope with these difficulties, we propose a fast and accurate extraction method of a vascular structure in 2D X-ray angiogram (XA) based on the vascular topology information of 3D computed tomography angiography (CTA) of the same patient. First, an initial vascular structure is robustly extracted based on vessel enhancement filtering. Then, 2D XA and 3D CTA are spatially aligned by 2D–3D registration. Finally, the 2D vascular structure is accurately reconstructed based on 3D vascular topology information by measuring the similarity of 2D and 3D vascular segments. Experimental results showed the fast and accurate extraction of a vascular structure using 10 images each of the left coronary artery (LCA) and right coronary artery (RCA). Our method can be used as a computer-aided technology for PCI. Our method can be applied to the study of other various types of vessels.