Assessment of fully automatic segmentation of pulmonary artery and aorta on noncontrast CT with optimal surface graph cuts
Zahra Sedghi Gamechi,Andres M. Arias‐Lorza,Zaigham Saghir,Daniel Bos,Marleen Bruijne
DOI: https://doi.org/10.1002/mp.15289
IF: 4.506
2021-10-29
Medical Physics
Abstract:PurposeAccurate segmentation of the pulmonary arteries and aorta is important due to the association of the diameter and the shape of these vessels with several cardiovascular diseases and with the risk of exacerbations and death in patients with Chronic Obstructive Pulmonary Disease (COPD). We propose a fully automatic method based on an optimal surface graph cut algorithm to quantify the full 3D shape and the diameters of the pulmonary arteries and aorta in non-contrast Computed Tomography (CT) scans. MethodsThe proposed algorithm first extracts seed points in the right and left pulmonary arteries, the pulmonary trunk, and the ascending and descending aorta by using multi-atlas registration. Subsequently, the centerlines of the pulmonary arteries and aorta are extracted by a minimum cost path tracking between the extracted seed points, with a cost based on a combination of lumen intensity similarity and multiscale medialness in 3 planes. The centerlines are refined by applying the path tracking algorithm to curved multi-planar reformatted scans and are then smoothed and dilated non-uniformly according to the extracted local vessel radius from the medialness filter. The resulting coarse estimates of the vessels are used as initialization for a graph-cut segmentation. Once the vessels are segmented, the diameters of the pulmonary artery (PA) and the ascending aorta (AA) and the ratio are automatically calculated both in a single axial slice and in a 10 mm volume around the automatically extracted pulmonary artery bifurcation level. The method is evaluated on non-contrast CT scans from the Danish Lung Cancer Screening Trial (DLCST). Segmentation accuracy is determined by comparing with manual annotations on 25 CT scans. Intra-class correlation (ICC) between manual and automatic diameters, both measured in axial slices at the pulmonary artery bifurcation level, is computed on an additional 200 CT scans. Repeatability of the automated 3D volumetric diameter and ratio calculations (perpendicular to the vessel axis) are evaluated on 118 scan-rescan pairs with an average in-between time of 3 months. ResultsWe obtained a Dice segmentation overlap of 0.94 ± 0.02 for pulmonary arteries and 0.96 ± 0.01 for the aorta, with a mean surface distance of 0.62 ± 0.33 mm and 0.43 ± 0.07 mm, respectively. ICC between manual and automatic in-slice diameter measures was 0.92 for PA, 0.97 for AA, and 0.90 for the ratio, and for automatic diameters in 3D volumes around the pulmonary artery bifurcation level between scan and rescan were 0.89, 0.95, and 0.86, respectively. ConclusionThe proposed automatic segmentation method can reliably extract diameters of the large arteries in non-ECG-gated non-contrast CT scans such as are acquired in lung cancer screening.This article is protected by copyright. All rights reserved
radiology, nuclear medicine & medical imaging