This retrospective multicenter study evaluated a fully automatic deep learning-based method for aortic segmentation and simultaneous diameter and volume measurements. The Augmented Radiology for Vascular Aneurysm (ARVA) tool from Incepto was used on 350 aortic CT Angiography scans (including preoperative and postoperative) from 216 patients treated at two different hospitals in France. ARVA’s ability to segment the aorta and measure maximum wall transverse diameters, and compute volumes, was compared with the measurements of six experts (ground truth) and thirteen clinicians. Its capacity to measure transverse outer-to-outer wall diameters showed no significant differences with clinicians (median absolute diameter difference of 1.6 mm in both comparisons) and remained within the clinically accepted error range (ARVA’s 85.5% compared to the clinicians’ 86.0%). Comprehensive analysis of the entire aorta volume achieved a median dice of 0.95 and median volume similarity of 0.98. The study demonstrates that the AI-driven solution is as accurate as clinicians for diameter measurements and performs reliable volume measurements, potentially contributing to shape the future of aortic surgery.
Read full study
Abstract
Objective
This study aims to evaluate a fully automatic deep learning-based method (Augmented Radiology for Vascular Aneurysm [ARVA]) for aortic segmentation and simultaneous diameter and volume measurements.
Methods
A clinical validation dataset was constructed from preoperative and postoperative aortic Computed Tomography Angiography (CTA) scans for assessing these functions. The dataset totalled 350 CTA scans from 216 patients treated at two different hospitals. ARVA’s ability to segment the aorta into seven morphologically based aortic segments and measure maximum outer-to-outer wall transverse diameters and compute volumes for each was compared with the measurements of six experts (ground truth) and thirteen clinicians.
Results
Ground truth (experts') measurements of diameters and volumes were manually performed for all aortic segments. The median absolute diameter difference between ground truth and ARVA was 1.6mm ([1.5, 1.7] 95% confidence interval (CI)), and 1.6mm ([1.6, 1.7] 95% CI) between ground truth and clinicians. ARVA produced measurements within the clinical acceptable range with a proportion of 85.5% ([83.5,86.3] 95% CI) compared to the clinicians' 86.0% ([83.9, 86.0] 95% CI). Median volume similarity error ranged from 0.93 to 0.95 in the main trunk and achieved 0.88 in the iliac arteries.
Conclusion
This study demonstrates the reliability of a fully automated AI-driven solution capable of quick aortic segmentation and analysis of both diameter and volume for each segment.
