Segmenting 3D branching tubular structures using cores

Abstract

Blood vessels and other anatomic objects in the human body can be described as trees of branching tubes. The focus of this paper is the extraction of the branching geometry in 3D, as well as the extraction of the tubes themselves via skeletons computed as cores. Cores are height ridges of a graded measure of medial strength called medialness, which measures how much a given location resembles the middle of an object as indicated by image intensities. The methods presented in this paper are evaluated on synthetic images of branching tubular objects as well as on blood vessels in head MR angiogram data. Results show impressive resistance to noise and the ability to detect branches spanning a variety of widths and branching angles.