Live-Vessel: Interactive Vascular Image Segmentation with Simultaneous Extraction of Optimal Medial and Boundary Paths

Vessel analysis is important for a wide range of clinical diagnoses and disease research such as diabetes and malignant brain tumours. Vessel segmentation is a crucial first step in such analysis but is often complicated by structural diversity and pathology. Existing automated techniques have mixed results and difficulties with non-idealities such as imaging artifacts, tiny vessel structures and regions with bifurcations. In this paper we propose Live-Vessel as a novel and intuitive semi-automatic vessel segmentation technique. Our contribution is two-fold. First, we extend the classic Live-Wire technique from userguided contours to user guided paths along vessel centrelines with automated boundary detection. Live-Vessel achieves this by globally optimizing vessel filter responses over both spatial (x, y) and non-spatial (radius) variables simultaneously. Secondly, our approach incorporates colour gradient and quaternion curvature image information in the segmentation process unlike the majority of current techniques which employ greyscale values or a single colour channel. We validate our method using real medical data from the Digital Retinal Images for Vessel Extraction (DRIVE) database. Quantitative results show that, on average, Live-Vessel resulted in a 8-fold reduction in overall manual segmentation task time, at a 95% accuracy level. We also compare Live-Vessel to state-of-the-art methods and highlight its important advantages in providing the correct topology of the vascular tree hierarchy as well as the associated vessel medial (skeletal) curves and thickness profiles without the need for subsequent error-prone post processing.