Meta-Analysis of Deep Learning Models for Aortic Dissection Segmentation and Diagnosis

48 studies on deep learning models for aortic dissection were analyzed, with 28 focused on segmentation and 20 on diagnosis. The review, authored by Yichen Zhao and last updated in April 2026, reports a mean Dice coefficient of 89.2% for false lumen segmentation and pooled sensitivity of 0.94 for CT-based diagnostic models. It concludes these models perform comparably to or better than clinicians, supporting their potential as clinical assistive tools.

Use Cases

• Benchmarking new segmentation models against reported performance metrics for false lumen, true lumen, and entire aorta segmentation.
• Evaluating the diagnostic accuracy of image-based AI models for aortic dissection based on pooled sensitivity and specificity results.
• Informing the design of clinical validation studies based on identified gaps in multicenter validation and workflow integration.

Strengths

• Includes performance metrics from 48 distinct studies, providing a broad evidence base.
• Reports specific quantitative results, such as a Dice coefficient of 91.7% for entire aorta segmentation and a pooled specificity of 0.95 for CTA-based models.
• Compares model performance directly to clinicians, with reported sensitivity of 0.79 and specificity of 0.95 for clinician performance.

Limitations

• The underlying dataset is a 23.3 MB DOCX file; the raw image data and model code from the reviewed studies are not included.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for direct data analysis.

Provenance

Source

Systematic review and meta-analysis compiled by author Yichen Zhao, published on figshare.

Collection Method

Comprehensive search across Cochrane Library, PubMed, Embase, and Web of Science up to November 3, 2024, with risk evaluation using QUADAS-2.

Time Range

Studies included up to November 3, 2024.

Freshness

Last updated 2026-04-14 04:21:47