Improved stent sharpness evaluation with super-resolution deep learning reconstruction in coronary CT angiography

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Cardiovascular Imaging

Background:

• Coronary CT angiography (CCTA) is crucial for visualizing coronary arteries.
• Image quality in CCTA can be limited by noise and blooming artifacts, especially around coronary stents.
• Conventional reconstruction methods like hybrid iterative reconstruction (HIR) and deep learning-based reconstruction (DLR) have limitations in artifact reduction and sharpness.

Purpose of the Study:

• To evaluate the effectiveness of super-resolution deep learning reconstruction (SR-DLR) in improving CCTA image quality.
• To compare SR-DLR against HIR and DLR in reducing blooming artifacts and enhancing stent visualization.
• To assess the impact of SR-DLR on image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and stent sharpness.

Main Methods:

• Retrospective analysis of 66 CCTA patient datasets.
• Quantitative assessment of image noise, SNR, and CNR.
• Quantification of stent sharpness using edge rise slope (ERS) and edge rise distance (ERD).
• Qualitative assessment using a 5-point scoring system for image quality, noise, vessel wall, and stent structure.

Main Results:

• SR-DLR demonstrated significantly lower image noise compared to HIR and DLR.
• SR-DLR achieved higher SNR and CNR values.
• SR-DLR significantly improved stent sharpness, evidenced by a lower mean ERD (0.70 mm) compared to HIR (1.13 mm) and DLR (0.85 mm).
• Qualitative scores for SR-DLR were superior across all assessed categories.

Conclusions:

• SR-DLR significantly enhances CCTA image quality by reducing noise and improving stent sharpness.
• SR-DLR offers a valuable advancement for coronary artery stent visualization without hardware constraints.
• The superior performance of SR-DLR in both quantitative and qualitative metrics highlights its potential clinical utility.