Abstract: Targeting at the possibly existing complex furcations and narrow tubes which are problematic in reconstructing vectorial surface of blood vessels,we introduce an efficient and reliable vector-based modeling method.First,the skeletal representation of vascular volume is extracted from its segmented volume and further converted into an adaptive one constituted by a number of single lines,whose optimal radii are computed simultaneously upon continuous center points.And then based on this adaptive skeleton and the optimal radii,we reconstruct tubular surface for each single line.And for each furcating part,we use our Delaunay triangulation-like method to merge the surfaces of its branching single lines into a closed one which meets the CGAL standard.Finally,the whole surface is further subdivided to provide an exquisite and accurate model.The Delaunay triangulation-like method takes full advantage of the eating-each-other relationships among these single lines with small computation overhead required.The experimental outcomes show that our method is effective and the resulted model is accurate,surpassing the pre-existing vector-based methods.