4:51 PM Abstract No. 235 - Clinical validation of a semi-automated software for maximal diameter measurements and follow-up of complex AAA geometries

Abstract

To compare automated measurement of abdominal aortic aneurysm (AAA) maximal diameter (D-max) orthogonal to luminal or thrombus centerline. The local ethic comity approved this retrospective study of 83 patients (74 men, 9 women; mean age, 75 years, range, 46-91 years) with AAA ≥ 35 mm followed by multidetector computer tomography (MDCT) with one scan pre-stenting and at least two scans three months post-EVAR. 3-D reconstruction of the AAA and its components (wall, lumen, thrombus and centerlines) was achieved with dedicated segmentation software. Performances of automatic calculation algorithms of D-max perpendicular to lumen or thrombus centerlines were then compared to manually measured D-max on double-oblique MPR projections. To assess agreement, mean difference, bland-altman plot, regression models and paired student’s t-test were performed at baseline, first follow-up (FU) scan after EVAR and on the progression interval (last FU scan versus baseline). There was no significant difference in D-max measurements between thrombus algorithm and manual method (at baseline, mean error: -0.07 ± 1.66 mm, p-value= 0.7; at FU, 0.24 ± 1.69 mm, p-value= 0.2) compared to the luminal algorithm (at baseline, mean error: -1.24 ± 2.01 mm, p-value < 0.01; at FU, -1.49 ± 3.30 mm, p-value < 0.01). The difference in D-max progression between manual and thrombus algorithm was not significant (p-value = 0.1). AAA thrombus segmentation is necessary to enable automated calculation of D-max and follow D-max of AAAs with complex geometries.