HIERARCHICAL SEARCH ALGORITHM FOR IDENTIFICATION OF CORONARY ARTERY CONTOURS IN DIGITAL ANGIOGRAMS.
Yuping Sun, Tongyan Wu, William J. Ohley, Devan Williams, A S Most · Journal of Media Literacy Education · 1987
A search algorithm to automatically determine the coronary artery contours in digital angiograms is presented. The search is hierarchical with the major arteries being identified first and then the side branches. The only operator intervention required is to define a start-of-search point and an initial search direction on the image. The search direction is adaptive, so no a priori assumption concerning the general orientation of the vessels is necessary. The algorithm is iterative over three steps, respectively, for the identification of 1) central artery lines; 2) artery edges, and 3) branch points. Output includes a line representation of the branching structure and a chain-connected contour delimiting the coronary artery segments. The algorithm was evaluated with both noise-added, phantom images and human coronary arteriograms. Preliminary results show that this algorithm can precisely segment the coronary artery branches with relatively complicated structures.