2019/08/14 by Jelmer M. Wolterink, Tim Leiner, Wolterink, Jelmer M. +3
Medicine · #Cardiac Imaging and Diagnostics #Cardiovascular Disease and Adiposity #Cerebrovascular and Carotid Artery Diseases #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1908.05343
openalex publication_date 2019/08/14 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28
Detection of coronary artery stenosis in coronary CT angiography (CCTA)\nrequires highly personalized surface meshes enclosing the coronary lumen. In\nthis work, we propose to use graph convolutional networks (GCNs) to predict the\nspatial location of vertices in a tubular surface mesh that segments the\ncoronary artery lumen. Predictions for individual vertex locations are based on\nlocal image features as well as on features of neighboring vertices in the mesh\ngraph. The method was trained and evaluated using the publicly available\nCoronary Artery Stenoses Detection and Quantification Evaluation Framework.\nSurface meshes enclosing the full coronary artery tree were automatically\nextracted. A quantitative evaluation on 78 coronary artery segments showed that\nthese meshes corresponded closely to reference annotations, with a Dice\nsimilarity coefficient of 0.75/0.73, a mean surface distance of 0.25/0.28 mm,\nand a Hausdorff distance of 1.53/1.86 mm in healthy/diseased vessel segments.\nThe results showed that inclusion of mesh information in a GCN improves\nsegmentation overlap and accuracy over a baseline model without interaction on\nthe mesh. The results indicate that GCNs allow efficient extraction of coronary\nartery surface meshes and that the use of GCNs leads to regular and more\naccurate meshes.\n