Stenosis Detection Algorithm for Coronary Angiograms

Abstract

Automatic detection of blocks in the angiographic images is a challenging task. The features such as contrast and gradient of the vessels and the background image are playing a vital role in the detection of the blocks in the X-Ray angiograms. Nowadays, doctors manually identify blocks in the coronary vessels. The automation tool is necessary to identify the blocks in the blood vessels of the heart to help the doctors in the diagnosing process. Spatiotemporal nature of the angiography sequences is used to isolate the coronary artery tree. The coronary artery segment is tracked and in each image frame by frame and the arterial width surface is detected. The stenosis identification is done by using coronary vessel surface’s persistent minima and blob analysis. The proposed method is experimented on 42 patients’ dataset. The performance of the proposed method was evaluated by comparing the blocks identified by the algorithm with the hand-labelled ground truth images given by the experts. The proposed method provides an accuracy of 95.5% on 42 patients with a total of 60 image runs.