Automated Assessment of Thoracic Aorta Morphology and Calcium Distribution

Abstract

Abstract-Calcifications are heterogeneously distributed within different vascular beds. The presence and extent of calcium can be measured from non-contrast CT images and has been associated with higher cardiovascular risks. Using heart scans, thoracic aorta calcium (TAC) can be simultaneously assessed, although the aortic arch is systematically excluded. In this work we developed an automated technique that estimated the thoracic aorta (TA) morphology in 3D and the associated calcifications. One hundred non-diabetic patients (male gender 50%, aging 55-to-60 y.o.) at intermediate risk were randomly selected. Extended non-contrast CT scans, that included the aortic arch, were conducted. A custom software segmented the thoracic aorta using an adaptive circle fitting method that estimated the aortic centerline and diameter in 140 points per patient. Traditional risk factors were similar between men and women (p=NS). As expected, the Framingham risk score was higher in men (p<0.001). All TA size measurement were larger in men, including mean diameter, length and volume (p<0.001). In shape variables, aortic arch width was wider in men (p<0.001), although the arch height and tortuosity were similar with respect to women (p=NS). We did not find differences in TAC distribution for the ascending, arch and descending segments. Calcifications were mostly concentrated in the aortic arch and the proximal descending portions. We conclude that, in spite of having larger and wider TAs, men and women had similar calcification distribution. Calcium appeared along the TA curvilinear portion that remains invisible in traditional coronary calcium examinations. Further analyses are required to evaluate if this extended TAC inspection could be used to revise its prognostic value in prospective studies.