Segmentation of Coronary Arteries in MSCT Datasets Using Livewire

Abstract

The goal of this project was the development of a fast segmentation algorithm for coronary arteries in MSCT datasets that simultaneously extracts the vascular tree and provides a rough estimate of the vessel's centerline. For that purpose, and due to considerations that will be presented later on, the segmentation technique for boundary extraction known as livewire was chosen as the basis. The project was developed in MeVisLab, a fast prototyping environment for medical image processing and visualization. The evaluation of the algorithm was carried out on a wide set of MSCT scans provided by the Institute of Diagnostic Radiology, University Hospital Zurich.