Artery skeleton extraction based on consistent curvature labeling

A method for detecting and extracting the skeletons of coronary arteries in coronary angiograms is proposed. The grayscale coronary angiograms are regarded as noisy sampling of the underlying continuous surface. After the application of Gaussian filtering to reduce noise, the topographic features of the smoothed image are detected based on first and second order image derivatives which characterize the local differential image structure. Then, the candidate arteries skeleton areas are detected based on the observation that arteries are smooth elongated objects having approximately a Gaussian smoothed semi-elliptical profile. False skeleton areas are eliminated through connected component analysis and the morphological operation of reconstruction. Experimental results on real digitized coronary angiograms are presented.