On the segmentation of vascular geometries from medical images

Abstract

SUMMARYA comprehensive analysis of vascular morphology and the application of generic models of vascular biomechanics to specific patients require the ability of extracting a geometrical representation of the vascular anatomy from medical images. Owing to the wide range of clinical manifestations of vascular disease and associated imaging modalities and protocols, several segmentation methods have been proposed over the last 20 years and are available in the literature. In this paper, we review the methods of segmentation of angiographic medical images and identify major advantages and disadvantages of stateof-the-art techniques. We further discuss the performance of some of the most popular intensity-based and gradient-based methods using a set of images of peripheral by-pass grafts acquired with magnetic resonance angiography (MRA). We then propose a threshold front method for the segmentation of MRA images and assess its performance using two anatomic scale replica models, reproducing a normal and a stenotic peripheral artery. The threshold front algorithm is a simple, fast and parameter-free (still adaptive) method achieving segmentation errors below pixel resolution.

Automated detection of intracranial aneurysms based on parent vessel 3D analysis

“…Because 3D-RA, CTA and MRA data provide vessel and aneurysm positions in cross-sectional images only, the extraction of 3D structures from 2D images is achieved through segmentation. A great deal of research has been carried out in developing algorithms for the segmentation of cerebral vasculature, including aneurysms, from MRA and CTA studies (Fridman et al, 2004;Hernandez and Frangi, 2007;Radaelli and Peiro, 2009). Vessels segmentation remains a challenging task and the research in this field remains active.…”

Image Segmentation for Cardiovascular Biomedical Applications at Different Scales

“…Each biomedical application imposes specific restrictions on both the input medical images and the output patient-specific model, and, therefore, calls for a specific class of 3D segmentation methods. Various medical image segmentation techniques have been developed [6][7][8].…”

Vessel segmentation in MRI using a variational image subtraction approach

“…There are several comprehensive reviews on vessel segmentation [7][8][9] and a vast number of research papers on the medical image segmentation problem [10][11][12][13][14][15], as either general segmentation methods or the segmentation of specific biological structures. Segmentation methods may be grouped into many different categories, such as histogram-and thresholding-based, clustering-based, edge-detection-based, region-growingbased, split-and-merge-based, graph-based, and partial differential equation-based [7].…”

“…By iteratively moving the snake points, it aims to minimize the total energy to enable the snake to fit the image features well. The total energy is the weighted sum of the energy of internal and external forces, such as the image gradient fitted to vessel structures [20]. The snake algorithm is extremely sensitive to the initialization and noise, as well as to the concave parts of image contour attracting the segmentation.…”