Search

OBJECTIVES: Establishing the reproducibility of expert-derived measurements on CTA exams of aortic dissection is clinically important and paramount for ground-truth determination for machine learning. METHODS: Four independent observers retrospectively evaluated CTA exams of 72 patients with uncomplicated Stanford type B aortic dissection and assessed the reproducibility of a recently proposed combination of four morphologic risk predictors (maximum aortic diameter, false lumen circumferential angle, false lumen outflow, and intercostal arteries). For the first inter-observer variability assessment, 47 CTA scans from one aortic center were evaluated by expert-observer 1 in an unconstrained clinical assessment without a standardized workflow and compared to a composite of three expert-observers (observers 2-4) using a standardized workflow. A second inter-observer variability assessment on 30 out of the 47 CTA scans compared observers 3 and 4 with a constrained, standardized workflow. A third inter-observer variability assessment was done after specialized training and tested between observers 3 and 4 in an external population of 25 CTA scans. Inter-observer agreement was assessed with intraclass correlation coefficients (ICCs) and Bland-Altman plots. RESULTS: Pre-training ICCs of the four morphologic features ranged from 0.04 (-0.05 to 0.13) to 0.68 (0.49-0.81) between observer 1 and observers 2-4 and from 0.50 (0.32-0.69) to 0.89 (0.78-0.95) between observers 3 and 4. ICCs improved after training ranging from 0.69 (0.52-0.87) to 0.97 (0.94-0.99), and Bland-Altman analysis showed decreased bias and limits of agreement. CONCLUSIONS: Manual morphologic feature measurements on CTA images can be optimized resulting in improved inter-observer reliability. This is essential for robust ground-truth determination for machine learning models. KEY POINTS: • Clinical fashion manual measurements of aortic CTA imaging features showed poor inter-observer reproducibility. • A standardized workflow with standardized training resulted in substantial improvements with excellent inter-observer reproducibility. • Robust ground truth labels obtained manually with excellent inter-observer reproducibility are key to develop reliable machine learning models.

Objective: Late adverse events (LAEs) are common among initially uncomplicated type B aortic dissection (uTBAD); however, identifying those patients at highest risk of LAEs remains a significant challenge. Early false lumen (FL) growth has been suggested to increase risk, but confident determination of growth is often hampered by error in two-dimensional clinical measurements. Semi-automated three-dimensional (3D) mapping of aortic growth, such as by vascular deformation mapping (VDM), can potentially overcome this limitation using computed tomography angiograms (CTA). We hypothesized that FL growth in the early pre-dissection phase by VDM can accurately predict LAEs., Methods: We performed a two-center retrospective study of patients with uTBAD, with paired CTAs in the acute (1-14 days) and subacute/early chronic (1-6 months) periods. VDM analysis was used to map 3D growth. Standard clinical CT measures (ie, aortic diameters, tear characteristics) were also collected. Multivariate analysis was conducted using a decision tree and Cox proportional hazards model. LAEs were defined as aneurysmal FL (>55 mm); rapid growth (>5 mm within 6 months); aorta-specific mortality, rupture, or re-dissection., Results: A total of 107 (69% male) patients with uTBAD initially met inclusion criteria with a median follow-up of 7.3 years (interquartile range [IQR], 4.7-9.9 years). LAEs occurred in 72 patients (67%) at 2.5 years (IQR, 0.7-4.8 years) after the initial event. A multivariate decision tree model identified VDM growth (>2.1 mm) and baseline diameter (>42.7 mm) as optimal predictors of LAEs (area under the receiver operating characteristic curve = 0.94), achieving an 87% accuracy (sensitivity of 93%, specificity of 76%) after leave-one-out validation. Guideline reported high-risk features were not significantly different between groups., Conclusions: Early growth of the FL in uTBAD was the best tested indicator for LAEs and improves upon the current gold-standard of baseline diameter in selecting patients for early prophylactic thoracic endovascular aortic repair., Competing Interests: Disclosures N.S.B. is entitled to royalties related to licensure of intellectual property from vascular deformation mapping (VDM) to Imbio Inc., (Copyright © 2024 Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.)

Patients with initially uncomplicated typeB aortic dissection (uTBAD) remain at high risk for developing late complications. Identification of morphologic features for improving risk stratification of these patients requires automated segmentation of computed tomography angiography (CTA) images. We developed three segmentation models utilizing a 3D residual U-Net for segmentation of the true lumen (TL), false lumen (FL), and false lumen thrombosis (FLT). Model 1 segments all labels at once, whereas model 2 segments them sequentially. Best results for TL and FL segmentation were achieved by model 2, with median (interquartiles) Dice similarity coefficients (DSC) of 0.85 (0.77-0.88) and 0.84 (0.82-0.87), respectively. For FLT segmentation, model 1 was superior to model 2, with median (interquartiles) DSCs of 0.63 (0.40-0.78). To purely test the performance of the network to segment FLT, a third model segmented FLT starting from the manually segmented FL, resulting in median (interquartiles) DSCs of 0.99 (0.98-0.99) and 0.85 (0.73-0.94) for patent FL and FLT, respectively. While the ambiguous appearance of FLT on imaging remains a significant limitation for accurate segmentation, our pipeline has the potential to help in segmentation of aortic lumina and thrombosis in uTBAD patients.Clinical relevance- Most predictors of aortic dissection (AD) degeneration are identified through anatomical modeling, which is currently prohibitive in clinical settings due to the timeintense human interaction. False lumen thrombosis, which often develops in patients with type B AD, has proven to show significant prognostic value for predicting late adverse events. Our automated segmentation algorithm offers the potential of personalized treatment for AD patients, leading to an increase in long-term survival.