A novel method for 3D reconstruction of coronary bifurcation using quantitative coronary angiography

Abstract

Many methods have been proposed for the 3-dimensional (3D) reconstruction of coronaries arteries by combing information from two or more X-ray views of the coronary tree, since the 2D representation of coronary lesion using X-ray coronary angiographies is limited. The aim of this study is to present a new semi-automated method for the accurate 3D reconstruction of coronary arterial bifurcations using X-ray coronary angiographic views (CA). X-ray angiography was acquired from seven patients, both pre and post angioplasty procedure, and their data were used for the 3D reconstruction methodology. The proposed approach consists of 3 steps. Initially, the 2D lumen borders and centerlines are detected. Then the 3D bifurcation path is extracted and the 3D lumen borders are reconstructed around the 3D bifurcation path and finally, the main and side segments are intersected in order to produce the finally model of the bifurcated artery. Considering the X-ray angiography as the gold standard, we validated the proposed method based on the 2D versus the 3D bifurcation segment model. More specifically, in the current dataset our results indicate excellent correlation with the 2D angiography: r = 0.98, p < 0.001; r = 0.97, p < 0.001; r = 0.94, p < 0.001 and r = 0.88, p < 0.001 for the Reference Lumen Diameter (RLD), Minimal Lumen Diameter (MLD), Degree of Stenosis (DS) and Lesion Length (LL), respectively. Moreover the mean values of the Hausdorff Distance and the Dice correlation between the 2D annotated borders and the forwardly projected borders are 0.3031 mm and 91% for the first CA while 0.3103 and 92% for the second one.