Automatic vessel tracking and segmentation using epicardial ultrasound in bypass surgery

Epicardial ultrasound has been suggested as an alternative approach for assessing the quality of coronary artery bypass graft anastomoses. Using automatic tracking and segmentation of the anastomotic vessel lumen in transverse epicardial ultrasound images it is possible to quantitatively assess the stenosis degree of surgical errors. We propose an automatic vessel tracking and segmentation framework that can detect, track, and segment vessels through ultrasound sequences with the purpose of enabling stenosis quantification. An average accuracy of 92.86% in detecting vessels was obtained 78.51% of the vessel segmentations were assessed as correct.