Automatic Segmentation of Coronary Arteries and Detection of Stenosis

Cardiovascular disease (CVD) is one of the most prevalent cause of death in many countries. Computed Tomography Angiography (CTA) is a widely used imaging modality to diagnose and treat different types of CVD. Diagnosing stenosis in coronary artery using CTA data sets is a tedious and time consuming task. To overcome this difficulty we propose an automated segmentation system for stenosis detection on 2D projection images. The segmentation of coronary artery is achieved by techniques such as image smoothing, vessel enhancement, localized threshold and connected component labeling. Further, stenosis, if present, is identified by finding the discontinuities in the vessel by centerline extraction, calculating the thickness and the intensity of the vessel. The objective of the proposed system is to reduce the number of false negative responses by finding out all suspected parts of the coronary arteries for detailed and final investigations by medical experts. The performance of the proposed system has been evaluated by comparing the outcome with the ground truth given by the experts and this provides an average recall measure of 97%.