Interobserver and intraobserver variability in measuring the tortuosity of the thoracic aorta on computed tomography.

Objective: The variability in measuring the tortuosity of the thoracic aorta has not been previously studied. This study evaluated the interobserver and intraobserver variability of major methods used for measuring the tortuosity of the thoracic aorta in patients with aortic arch or descending thoracic aortic aneurysm.

Methods: This retrospective study enrolled 66 patients with aortic arch or descending thoracic aortic aneurysm who had undergone thoracic endovascular aortic repair. Two radiologists used preoperative computed tomography images to measure the tortuosity of the thoracic aorta at multiple segments by using the fitting circle diameter, tortuosity index, and centerline angle methods; these measurements were repeated after an interval of >28 days. The variability of the methods was analyzed for interobserver and intraobserver reliability and agreement. The estimated intraclass correlation coefficient (ICC) was used to analyze the reliability. The Bland-Altman plot was used to analyze the interobserver and intraobserver agreement. The association between aortic characteristics, including calcification, luminal irregularity, shape, and diameter, and the variability of the measurements was also analyzed.

Results: The interobserver ICC estimates for the tortuosity index at multiple aortic segments, centerline angle methods at the supra-aortic branch orifices, and fitting circle diameter on the greater and lesser curvature sides were 0.97 to 0.98, 0.39 to 0.75, and 0.82 to 0.84, respectively. The corresponding intraobserver ICC estimates were 0.98 to 1.00, 0.44 to 0.75, and 0.82 to 0.85, respectively. In the agreement analysis, the 95% limits of agreement for the tortuosity index, centerline angle, and fitting circle diameter were -5.5% to 5.6%, -10.9% to 10.9%, and -18.0% to 24.0%, respectively. The tortuosity index had the highest ICC estimate and narrowest 99.5% limits of agreement of the three methods. Aortic characteristics, including calcification, grade of atheroma, aneurysm shape, and diameter, were not associated with the variability of the tortuosity index method in the thoracic aorta.

Conclusions: The tortuosity index method has low interobserver and intraobserver variability in measuring the tortuosity of the thoracic aorta in patients with thoracic aortic aneurysm. The characteristics of the aorta and aneurysm are not associated with the interobserver or intraobserver variability of the tortuosity index.