What is the clinical evidence from this study?

Study design: Meta-Analysis. Population: Aortic dissection (n=31198). Intervention: Image-based deep learning models vs. Reference standard / Clinicians. Primary outcome: Diagnostic accuracy for aortic dissection (DOR 117, 95% CI 70-196).

April 14, 2026Open Access

Performance of image-based deep learning models for aortic dissection segmentation and diagnosis: a systematic review and meta-analysis

Key Result

Image-based deep learning models demonstrated high diagnostic accuracy for aortic dissection, achieving a pooled sensitivity of 0.91, specificity of 0.92, and a diagnostic odds ratio of 117.

Study Design

Type

Meta-Analysis (n=31,198)

Structured PICO

Does image-based deep learning accurately segment and diagnose aortic dissection compared to reference standards and clinicians?

Population

48 studies (28 on segmentation tasks involving 1,810 aortic dissection patients with 9,488 images; 20 on diagnostic tasks involving 29,388 patients with 92,319 images, of whom 2,560 had aortic dissection)

Intervention

Image-based deep learning models (using CT, CTA, ECG, etc.) for aortic dissection segmentation and diagnosis

Comparator

Reference standard (ground truth) and clinician diagnosis

Outcome

Diagnostic accuracy metrics including Dice coefficient for segmentation, and sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and area under the summary receiver operating characteristic curve (AUSROC) for diagnosissurrogate

Image-based deep learning models demonstrate high accuracy for aortic dissection segmentation and diagnosis, performing comparably to or better than clinicians, supporting their potential as clinical assistive tools.

Main Result

Effect estimate: DOR 117 (95% CI 70-196)

Limitations

Absence of standard errors or standard deviations for Dice coefficients in most segmentation studies prevented conventional meta-analysis.
Scarcity of studies on thrombus segmentation.
Heterogeneity in validation methods, with most relying on internal validation (random splitting or cross-validation) and only five using external validation.
Limited sample sizes and lack of diversity in training data due to the low incidence of aortic dissection.
Variations in image acquisition protocols and scanning equipment across institutions.
Potential publication bias in certain subgroup analyses.

Abstract

Objective: This meta-analysis was conducted to systematically evaluate the accuracy of image-based deep learning models for aortic dissection segmentation and diagnosis, aiming to provide an evidence base for developing intelligent detection tools. Methods: A comprehensive search was performed across the Cochrane Library, PubMed, Embase, and Web of Science to identify studies on the effectiveness of deep learning in aortic dissection segmentation or diagnosis up to November 3, 2024. Risk evaluation was carried out with the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2). Results: A total of 48 studies on deep learning models for aortic dissection segmentation or diagnostic tasks were included, with 28 on segmentation tasks and 20 on diagnostic tasks. For segmentation tasks, the mean Dice coefficient was 89.2% ± 4.4% for false lumen segmentation, 90.8% ± 3.4% for true lumen segmentation, and 91.7% ± 6.1% for entire aorta segmentation. For diagnostic tasks, computed tomography (CT) based deep learning showed pooled sensitivity and specificity of 0.94 95% confidence interval (CI): 0.89-0.96 and 0.92 (95% CI: 0.88-0.95), respectively. In terms of electrocardiogram-based deep learning, the pooled sensitivity and specificity were 0.85 (95% CI: 0.79-0.89) and 0.90 (95% CI: 0.87-0.92), respectively. Regarding the computed tomography angiography (CTA) based deep learning, the pooled sensitivity and specificity were 0.94 (95% CI: 0.90-0.96) and 0.95 (95% CI: 0.91-0.98), respectively. Additionally, some studies compared the diagnostic performance of deep learning with that of clinicians. The pooled sensitivity and specificity were 0.79 (95% CI: 0.65-0.89) and 0.95 (95% CI: 0.88-0.94), respectively. Conclusions: Image-based deep learning models demonstrated high accuracy for aortic dissection segmentation and diagnosis. They performed comparably to or better than clinicians. These findings support their potential as clinical assistive tools. Future work should prioritize multicenter validation, seamless integration of these models into clinical workflows, and enhancement of model generalizability to facilitate broader clinical adoption. Systematic Review Registration: https://www.crd.york.ac.uk/PROSPERO/view/CRD42024619403, identifier CRD42024619403.

Mark Helpful

Bookmark

Relay

View Full Paper