Calculation of the aortic arch angles from three-dimensional reconstructions of computed tomography scans: comparison between an automated program and visual assessment.

POLITI, MARIA TERESA; CLAUDIA CAPURRO; JUAN FERNADEZ; JOSE-MARIA FULLANA; GAUDRIC, JULIEN; CARRE, EMMANUELLE

COMPUTERS IN BIOLOGY AND MEDICINE

PERGAMON-ELSEVIER SCIENCE LTD

Año: 2019

0010-4825

Background: Three dimensional (3D) reconstructions of computed tomography (CT) scans are currently the preferred imaging technique for evaluating aortic morphology. However, the adequate assessment of the angles of the aorta continues to represent a major difficulty. Aortic angulations are an aspect of the vessel?s anatomy which is often overlooked which possibly contributeing to the generation of endoleaks during endovascular procedures. We developed a new program to identify the location of the aortic points with maximum curvature, allowing to automatically calculate main angulations of the aortic arch that could have potential clinical value for discriminating patients unfit for endovascular treatment in case of thoracic aneurysms.Methods: This is a cross-sectional study of a convenience sample of subjects with multislice CT angiography scans of the thoracic aorta from the institutional imaging database. The center lumen line (CLL) of the aorta was determined semi-automatically using Endosize software. The points of maximum curvature of the CLL were determined twice: by visual assessment by two physicians and by a custom program we developed. These points were then used to calculate four indicators of aortic curvature: the aortic angles, mean curvature, tortuosity, and curvature length index (CLI).Results: The study enrolled 9 subjects: 4 with thoracic aneurysms and 5 with normal aortas. The inter-observer, inter-method and intra-observer correlation and agreement coefficients for each of the 3D spatial coordinates of the maximum curvature points were appropriate. There was no association between the presence of aneurysms and the median value of aortic angles (p=0.7783), the mean curvature (p=0.1089) or the aortic tortuosity (p=0.0500). However the CLI was significantly higher in subjects with aneurysms as compared to controls (34.4 [32.4-41.2] vs 26.9 [25.6-27.8]; p=0.0139).Conclusion: This proposed process identifies points of maximum curvature with good correlation with visual assessment and can also automatically calculate aortic angles, mean curvature, tortuosity and CLI. The CLI is a potential candidate to automatically discriminate CT scans of thoracic aneurysms from healthy volunteers.