Automated coronary artery extraction using watersheds

An algorithm for the automated extraction of the skeletons and borders of coronary arteries in digitized angiograms is proposed. Initially, the approximate skeleton and borders of the coronary artery tree are extracted through an artery tracking method based on circular template analysis. The skeleton and borders of each artery segment are used for constructing its enclosing area where the defined skeleton and border curves are considered as markers. Using the marked artery segment enclosing area (ASEA), an artery gradient image is constructed where all pixels inside the ASEA, except skeleton ones, are assigned the gradient magnitude of the original image. The markers of the artery gradient image are imposed as its unique regional minima by the homotopy modification method. Then, the watershed transform is applied for extracting the artery segment borders. Experimental results using digitized coronary angiograms are presented