Influence of computed tomography angiography reconstruction software on anatomic measurements and endograft component selection for endovascular abdominal aortic aneurysm repair.

ABSTRACT

OBJECTIVE: Three-dimensional (3D) centerline reconstruction of computed tomography angiography (CTA) images permits detailed anatomic characterization of abdominal aortic aneurysms and facilitates planning of endovascular repair. Although several programs for 3D CTA reconstruction and measurement are available, direct comparisons have not been published, and reliability between software platforms has not been characterized. We evaluated agreement between anatomic measurements obtained from 3D CTA reconstructions using three commercially available software programs and characterized concordance between the programs for endograft component selection. METHODS: Images from 92 CTA studies performed before abdominal aortic aneurysm repair were reconstructed and measured using three different software programs independent reconstruction with proprietary software (Preview; M2S Inc, Lebanon, NH), surgeon-based reconstruction with proprietary software (AquariusNet Thin Client; TeraRecon Inc, San Mateo, Calif), and surgeon-based reconstruction with open-source software (Osirix MD; Pixmeo, Geneva, Switzerland). Agreement between outer wall diameter and length measurements obtained from centerline reconstructions created with each program was evaluated using scatter plots, intraclass correlation coefficients, and Bland-Altman plots. Concordance between aortic and iliac endograft component diameters selected from measurements with each program based on published instructions for use was examined using weighted κ statistics. RESULTS: Diameter measurements were generally similar between programs. Mean diameters at all locations were within ≤ 1 mm of one another, and mean length measurements were within ≤ 10 mm of one another for all pairwise comparisons. Intraclass correlations coefficients between programs for diameter measurements were comparable between programs (≥ 0.82 for all diameter comparisons and ≥ 0.88 for all length comparisons) and indicated good correlation. Pair-wise comparisons indicated similar rates of identical and adjacent size endograft component selection without an obvious trend toward superior agreement for any two programs. Rates of identical proximal endograft diameter selection ranged from 46% to 59%, whereas 89% to 100% of proximal endograft diameters selected between programs were within one adjacent (smaller or larger) size of each other. For iliac endograft selection, rates of identical component diameter selection between programs ranged from 36% to 69%, and 58% to 99% of selected iliac endograft diameters were within one adjacent size. CONCLUSIONS: Outer wall diameter and centerline length measurements obtained from 3D CTA reconstructions demonstrated good correlation between imaging analysis software programs, and graft diameter selections based on these measurements were reasonably similar. Comparable 3D CTA reconstruction measurements can be generated from independent and surgeon-based approaches using proprietary and open-source software, and the selection of a method to interpret images for endograft planning can be individualized according to operator experience and available resources while retaining sufficient accuracy.