Semi-automated detection and quantification of aortic atheromas from three-dimensional transesophageal echocardiography

Quantification of descending aorta atheromas (AT) with 2D transesophageal echocardiography (TEE) is time-consuming and underestimates plaque burden, while 3D TEE can provide the basis for analysis of AT 3D dimensions and volume. We developed a novel semi-automated algorithm to detect aortic AT in 3D TEE, to quantify their severity and volume and tested its accuracy compared to an expert gold standard. The 3D TEE datasets acquired from 58 consecutive patients were analyzed. Results showed a good accuracy in comparison with expert analysis, with an agreement of 95% in absolute presence/absence of AT per patient, 89% in AT number and location and 85% in patient risk due to AT severity classification. AT volumes were highly correlated (R
²
=0.98), with a slight underestimation (9%) compared to manual measurements. The proposed semi-automated algorithm is rapid, feasible and accurate in analyzing AT from 3D TEE descending aorta datasets.