Carotid stenosis assessment with multi-detector CT angiography: comparison between manual and automatic segmentation methods

Chengcheng Zhu,Andrew J. Patterson,Owen M. Thomas,Umar Sadat,Martin J. Graves,Jonathan H. Gillard
DOI: https://doi.org/10.1007/s10554-012-0148-8
2012-11-08
The International Journal of Cardiovascular Imaging
Abstract:Luminal stenosis is used for selecting the optimal management strategy for patients with carotid artery disease. The aim of this study is to evaluate the reproducibility of carotid stenosis quantification using manual and automated segmentation methods using submillimeter through-plane resolution Multi-Detector CT angiography (MDCTA). 35 patients having carotid artery disease with >30 % luminal stenosis as identified by carotid duplex imaging underwent contrast enhanced MDCTA. Two experienced CT readers quantified carotid stenosis from axial source images, reconstructed maximum intensity projection (MIP) and 3D-carotid geometry which was automatically segmented by an open-source toolkit (Vascular Modelling Toolkit, VMTK) using NASCET criteria. Good agreement among the measurement using axial images, MIP and automatic segmentation was observed. Automatic segmentation methods show better inter-observer agreement between the readers (intra-class correlation coefficient (ICC): 0.99 for diameter stenosis measurement) than manual measurement of axial (ICC = 0.82) and MIP (ICC = 0.86) images. Carotid stenosis quantification using an automatic segmentation method has higher reproducibility compared with manual methods.
What problem does this paper attempt to address?