Carotid Artery Wall Segmentation by Coupled Surface Graph Cuts

Abstract

Abstract. We present a three-dimensional coupled surface graph cut algorithm for carotid wall segmentation from Magnetic Resonance Imaging (MRI). Using cost functions that highlight both inner and outer vessel wall borders, the method combines the search for both borders into a single graph cut. Our approach requires little user interaction and can robustly segment the bifurcation section of the carotid artery. Experiments on 32 carotid arteries from 16 patients show good agreement between manual segmentation performed by an expert and our method. The mean relative area of overlap is more than 85% for the lumen and outer vessel wall. In addition, differences in measured wall thickness with respect to the manual annotations were smaller than the image pixel size.

High shear stress relates to intraplaque haemorrhage in asymptomatic carotid plaques

“…The BioStress tool is a matlab network that streamlines the various steps that need to be taken to calculate shear stress using computational fluid dynamics (CFD). As an input for this tool the 3D lumen surface was segmented using an in-house developed validated automatic segmentation tool, detecting lumen and outer wall surface in the black-blood MRI and PDw-EPI images [23]. The segmentations were obtained from all available MRI cross sections covering the largest possible distance along the vessel within the margins of acceptable image quality.…”

Automatic segmentation of MR depicted carotid arterial boundary based on local priors and constrained global optimisation

“…In previous work, Ukwatta et al [23] and Arias et al [24] also made use of graph cut optimisation to segment the carotid artery wall. Ukwatta et al developed a continuous max‐flow‐based algorithm to enforce the inter‐surface order constraint to improve the segment accuracy.…”

“…However, if the OB alienates from its regular shape due to the existence of plaques, the proposed model can accumulate evidence to guide the segmentation of OB to its accurate location. Arias et al [24] defined a flow line that keeps the LBs and OBs as non‐intersecting. They further introduced the inter edges to penalise the abnormal segments, such as LBs and OBs locate too far, or OBs are inside LBs.…”

“…However, deformable models used in these studies were prone to converge to local minima, and consequently, a good initialisation should be provided manually by users. Recently, a graph cut optimisation is employed to perform the carotid artery wall segmentation [23, 24]. Arias et al [24] added the inner and outer borders of the carotid artery into the cost functions of graph‐cut optimisation to obtain a global solution.…”

“…Recently, a graph cut optimisation is employed to perform the carotid artery wall segmentation [23, 24]. Arias et al [24] added the inner and outer borders of the carotid artery into the cost functions of graph‐cut optimisation to obtain a global solution. However, the segmented results were not smoothed and had relatively large deviations when compared with the manual delineations by experts.…”