Proximal thoracic aortic diameter measurements at CT: repeatability and reproducibility according to measurement method.

To determine the variability in CT measurements of proximal thoracic aortic diameters obtained using double-oblique short axis and semiautomatic centerline analysis techniques. Institutional review board approval, with waiver of informed consent, was obtained for this HIPAA-compliant, retrospective study. Cardiac gated thoracic aortic CT scans were evaluated in 25 patients. Maximum aortic diameter measurements at the annulus, sinuses, sinotubular junction and ascending aorta were generated using double-oblique short axis and semiautomatic centerline analysis techniques. Intraobserver and interobserver variability and variability between techniques were assessed using the Wilcoxon signed rank test, Spearman's correlation coefficients and Bland-Altman plots. Mean intraobserver diameter differences using double oblique views ranged from -0.3 to 0.6 mm. The 95 % confidence interval for difference in diameters was ±2.4 to ±5.1 mm for radiologist #1 and ±2.6 to ±5.2 mm for radiologist #2, depending on location. Mean intraobserver diameter differences using centerline analysis ranged from 0.2 to 2.3 mm, and the 95 % confidence interval for difference in diameters was ±2.0 to ±4.6 mm, depending on location. Significant interobserver differences were seen for both double oblique views and centerline analysis. Measurements obtained using the two methods were strongly correlated (r = 0.81-0.99), although they were consistently larger using centerline analysis (95 % confidence interval, ±1.8 to ±3.2 mm). Although measurement variability of the proximal thoracic aorta was generally low using double oblique and centerline analysis techniques, differences of up to approximately 5 mm in diameter occurred within the 95 % confidence interval. Neither technique was clearly more reliable than the other.