Automated Detection and Differentiation of Stanford Type A and Type B Aortic Dissections in CTA Scans Using Deep
Hung-Hsien Liu1, Chun-Bi Chang2, Yi-Sa Chen1
1Department of Medical Imaging and Intervention, New Taipei City Municipal Tucheng Hospital, New Taipei City 236043, Taiwan.
Diagnostics (Basel, Switzerland)
|January 11, 2025
Summary
A new deep learning model accurately detects type A aortic dissection (AD) from CT scans, aiding rapid diagnosis in emergencies. This AI tool differentiates type A AD from type B AD and normal cases, potentially reducing mortality rates.
Area of Science:
- Artificial intelligence in medical imaging
- Cardiovascular radiology
- Deep learning for disease detection
Background:
- Aortic dissection (AD) is a life-threatening condition requiring rapid diagnosis.
- Accurate differentiation between Type A and Type B AD is critical for treatment.
- Current diagnostic methods can be time-consuming, impacting emergency care.
Purpose of the Study:
- To develop and validate a deep learning model for automatic detection of Type A AD.
- To differentiate Type A AD from normal cases and Type B AD.
- To assess the model's performance and processing time on CT angiography scans.
Main Methods:
- Retrospective study using aortic computed tomography angiography (CTA) scans from 498 patients.
- Development of a two-component deep learning model: aorta detection and dissection classification.
- Validation on an independent test set of 316 patients.
Main Results:
- The model achieved high sensitivity and specificity for Type A AD (0.969, 0.982) and Type B AD (0.946, 0.996).
- Excellent performance was also observed for normal cases (0.988, 1.000).
- Average processing time per CTA scan was rapid at 7.9 ± 2.8 seconds.
Conclusions:
- The deep learning model accurately and swiftly detects Type A AD.
- It shows potential as an imaging triage tool in emergency settings.
- Facilitates early intervention and surgery, aiming to decrease mortality rates for Type A AD patients.
