A Novel Hybrid Method for the Segmentation of the Coronary Artery Tree in 2D Angiograms

Nowadays, medical diagnostics using images have considerable importance in many areas of medicine. Specifically, diagnoses of cardiac arteries can be performed by means of digital images. Usually, this diagnostic is aided by computational tools. Generally, automated tools designed to aid in coronary heart diseases diagnosis require the coronary artery tree segmentation. This work presents a method for a semiautomatic segmentation of the coronary artery tree in 2D angiograms. In other to achieve that, a hybrid algorithm based on region growing and differential geometry is proposed. For the validation of our proposal, some objective and quantitative metrics are defined allowing us to compare our method with another one proposed in the literature. From the experiments, we observe that, in average, the proposed method here identifies about 90% of the coronary artery tree while the method proposed by Schrijver&Slump (2002) identifies about 80%.