Robust Semi-automated Path Extraction for Visualising Stenosis of the Coronary Arteries

Abstract

Computed tomography angiography (CTA) is useful for diagnosing and planning treatment of heart disease. However, contrast agent in surrounding structures (such as the aorta and left ventricle) makes 3-D visualisation of the coronary arteries difficult. This paper presents a composite method employing segmentation and volume rendering to overcome this issue. A key contribution is a novel Fast Marching minimal path cost function for vessel centreline extraction. The resultant centreline is used to compute a measure of vessel lumen, which indicates the degree of stenosis (narrowing of a vessel). Two volume visualisation techniques are presented which utilise the segmented arteries and lumen measure. The system is evaluated and demonstrated using synthetic and clinically obtained datasets.