Skip to main content
2026 SVS Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Presentation Icons
Ticket Required
Ticket Required
AI Presentation
AI Presentation
Hot Topic
Hot Topic
Taste of Boston
Taste of Boston

International Plenary

IP11 - Effect of artificial intelligence-powered reconstruction to optimize image quality of completion cone-beam computed tomography assessment of fenestrated-branched endovascular aortic repair

Wednesday, June 10, 2026
2:26 PM - 2:31 PM ET
Location: Hynes CC, Third Level, Room 306
Objectives: To evaluate and compare a novel artificial intelligence-powered (AIP) reconstruction algorithm to optimize image quality and reduce streak artifacts linked to blood flow pulsatility, including metallic artifacts, during completion cone-beam computed tomography (CBCT) assessment of fenestrated-branched endovascular aortic repair (FB-EVAR) for complex aortic aneurysms.

Methods: This is a multicenter, blinded, cross-sectional cohort study of patients treated by FB-EVAR with completion intraoperative CBCT at two aortic centers between 2023-2025. Each CBCT acquisition (with/without contrast) was reconstructed using both conventional and AIP algorithms. Image quality was retrospectively assessed by three independent experienced reviewers. In Phase 1, blinded evaluations were performed using a modified European Quality Criteria framework assessing image sharpness, streak artifacts, and diagnostic confidence, with repeat assessment after a 30-day interval to minimize recall bias. In Phase 2, a non-blinded side-by-side comparison of AIP and conventional reconstructions was conducted and classified as “better”, “no difference”, or “inferior”.

Results: A total of 100 patients (78% male; mean age 70 ± 9.5 years) with 181 CBCT acquisitions were analyzed, including 110 non-contrast and 71 contrast-enhanced CBCTs. In Phase 1, AIP reconstruction demonstrated consistently higher ratings across all image-quality domains. Compared with conventional reconstruction, AIP algorithm was associated with significantly higher rating of images as very sharp (81.2% vs 29.9%, p<.001) and with no or mild streak artifacts (98.1% vs 43.8%, p<.001; Figure 1). Diagnostic confidence was also higher with AIP reconstruction (97.7% vs 62.8%, p<.001). In Phase 2, AIP reconstruction was rated as providing superior diagnostic image quality in 94.2% of cases, with no difference in 5.8% and no cases rated as inferior (χ² = 422.1, p<.001). Among cases in which AIP images were rated as “better”, improved assessment of clinical findings typically prompting immediate revision was reported in 72.2% of patients. Non-contrast CBCTs demonstrated a higher proportion of AIP reconstructions rated as “better quality” compared with contrast-enhanced CBCT (96.9% vs 90.1%; p<.001), whereas “no difference” ratings were more frequent with contrast-enhanced acquisitions (9.9% vs 3.1%; p<.001) (Figure 2).

Conclusions: AI-powered reconstruction significantly improves diagnostic image quality and decreases streak artifact of completion CBCT assessment of FB-EVAR, enhancing evaluation of image findings typically prompting intraoperative revision. The benefit is most noticeable for non-contrast CBCT. The significance of image quality improvement on immediate diagnosis and revision of clinically significant technical defects needs to be further determined. These results support the integration of AIP reconstruction into intraoperative CBCT workflows for complex aortic endovascular repair.