Ridge-based automatic vascular centerline tracking in X-ray angiographic images

The extraction of vascular trees is very important for quantitative analysis of vascular structures. As angiographic image is the integration of X-ray through the whole body anatomy on the image plane, vascular structure loses most 3-D topological information. Hence, accurate vascular structure detection is of great help for clinical diagnosis. In this paper, a fully automatic vascular centerline extraction method is proposed. A self-adaptive morphological operator is combined with a multi-scale enhancement filter to enhance tubular-like structures. Then, points with local maximum intensity are extracted as seed points, while the initial track directions are determined by detecting prominent ridge points in the predefined range. By iteratively searching the connected ridge points, the centerlines are gradually extracted by connecting the ridge points. By statistically counting of connected components, fake connections are efficiently removed. And bifurcation points are discriminated from centerline skeletons by determining the connections of each centerline point. Our approach is automatic completely. Experimental results show that the proposed algorithm is very effective for the extraction of centerlines from angiographic images.