Segmentation method of degree-based transition region extraction for coronary angiograms

Abstract

Aiming at the complex background of coronary angiograms, weak contrast between the coronary arteries and the background, a new segmentation method based on transition region extraction of degree is proposed. Firstly, the paper analyzes the characteristic of coronary angiograms. Secondly, 6 different Gaussian matched templates are used to enhance the coronary angiograms in order to remove the background. Finally, the coronary arteries are obtained with the method of degree-based transition region extraction. The experiments indicate that the proposed method outperforms the segmentation method based on not only top-hat but also Gaussian filter about the small vessels extraction, connectivity and effectiveness. The method is indeed valuable for diagnosis and the quantitative analysis of coronary arteries.