The standard for tracking aneurysm sac regression after EVAR is based on a single number, maximum diameter, measured by hand on serial CT scans. It’s a strong marker of repair success, but it’s also subjective, inconsistently measured, and it captures only one dimension of a three-dimensional problem. Volumetric measurement promises a more sensitive picture of sac behavior, but until now the challenge has been doing it accurately, consistently, and fast enough to be practical. 

A new study presented at the 2026 Vascular Annual Meeting (VAM26) in Boston puts Cydar Maps to that test. 

Researchers from the University of South Florida Health Division of Vascular Surgery and  Cydar Medical evaluated the accuracy of Automatic Volume Assessment (AVA) — a feature of Cydar Maps, an FDA-cleared AI driven software device used for EVAR planning, intraoperative guidance, and postoperative surveillance. 

AVA is a deep-learning segmentation model trained on more than 2,000 CT scans from diverse patient populations. It automatically segments the full aortic anatomy and reports sac volume from the infrarenal aorta plus both common iliac arteries — the regions most relevant to EVAR surveillance. 

To gauge accuracy, the teams compared AVA’s automated volumes against manual expert CT scan segmentations (n=268). They then assessed AVA across a far larger real-world set: more than 6,000 processed patient “Maps” drawn from CT aortograms at over 50 US and European hospitals since 2015. 

The headline results point to accurate volumetric measurements using AVA: 

• High agreement with manual segmentation. Median Dice score of 0.95 (a measure of overlap between automated and manual contours, where 1.0 is perfect). 

• Low volume error. Median absolute volume error of just 6.2 ml. 

• Few outliers. Only 2.6% of patients (7 of 268) showed significant error above 50 ml. 

• Fast. After upload, volume analysis took under 10 seconds per scan. 

The real-world surveillance data was equally telling. Among patients with paired follow-up scans (3–6 months and again at 6–24 months), just over half showed sac regression (50.3% overall; 60.6% at the host institution), with median regression of 11 ml overall and 23 ml locally. This is the kind of change a volume-based view surfaces more readily than diameter alone. 

Current surveillance guidelines flag a sac for reintervention workup at ≥5 mm of diameter growth, but no equivalent threshold has been established for volume. This data suggests AVA delivers accurate, automated, and reproducible volume measurement consistently and quickly across thousands of scans in real clinical practice. 

The authors note the need to understand which imaging factors predict the rare large segmentation errors, and integrate AI-derived volumes directly into CT reporting workflows so the numbers reach clinicians at the point of decision. 

The takeaway for vascular teams is straightforward: the technology to make volumetric surveillance routine is here, validated, and already integrated within Cydar Maps. 

Learn more about AVA

Talk to our team about a demo and see how AVA fits in your Endovascular program.