The results of the Bland-Altman analysis showed no significant mean bias between the manual and automatic measurements of aortic diameters, with limits of agreement of ± 3 mm, which is clinically acceptable. These findings suggest that the AI tool performs at a comparable level of accuracy to manual measurements performed by experienced surgeons. This is a promising outcome, as it demonstrates the potential of AI to provide reliable, automated measurements that could be used in clinical practice.

Additionally, the Cohen’s Kappa values for the categorical data showed good agreement between the manual and automated methods. Specifically, the Kappa for the quantification of wall thrombus in the proximal sealing zone was 0.77, indicating substantial agreement, while the Kappa for the quantification of calcification in the distal sealing zone was 0.65, suggesting moderate agreement. These results further support the accuracy and utility of the AI tool in assessing key features that influence EVAR planning.

One of the most notables advantages of the AI tool is its ability to save time. On average, the automated extraction process took 8.8 ± 2.2 minutes per scan, compared to 24.9 ± 5.3 minutes for the manual process. The difference in time was statistically significant (p < 0.001), indicating that the AI tool is much faster than the traditional manual method. This time savings could significantly improve workflow efficiency, allowing surgeons to allocate more time to other aspects of patient care.