Cooperative carotid artery centerline extraction in MRI

Abstract

Centerline extraction of the carotid artery in MRI is important to analyze the artery geometry and to provide input for further processing such as registration and segmentation. The centerline of the artery bifurcation is often extracted by means of two independent minimum cost paths ranging from the common to the internal and the external carotid artery. Often the cost is not well defined at the artery bifurcation, leading to centerline errors. To solve this problem, we developed a method to cooperatively extract both centerlines, where in the cost to extract each centerline, we integrate a constraint region derived from the estimated position of the neighbor centerline. This method avoids that both centerlines follow the same cheapest path after the bifurcation, which is a common error when the paths are extracted independently. We show that this method results in less error compared to extracting them independently: 10 failed centerlines Vs. 3 failures in a data set of 161 arteries with manual annotations. Additionally, we show that the new method improves the non-cooperative approach in 28 cases (p < 0.0001) in a data set of 3,904 arteries.

UR-CarA-Net: A Cascaded Framework With Uncertainty Regularization for Automated Segmentation of Carotid Arteries on Black Blood MR Images

“…In the last decades, several research groups reported on various methods to enable automated plaque characterization on multi-contrast MRI by segmenting plaque components. Computer vision approaches such as shape fitting, active contours, and level sets, in combination with simple machine learning methods such as classification and clustering, were attempted early on [9,15,16,17,18,19,20,8,21,10,11,12,13,14]. In more recent years, convolutional neural networks (CNN), including U-Net, have gained increasing attention [22,23,24,25,26,27,28].…”

“…Sometimes even additional delineation of a muscle region is needed for intensity re-scaling [9,15,16]. While there are some studies where the CA was located using the lumen seed points in the distal slices alone, or in different CA branches [18,11,19,12,13], user interaction is still necessary, and no studies show robustness to seed positioning. A couple of publications report on automated CA localization, but the detection area is limited to manually selected slices [10].…”