Incremental Image Noise Reduction in Coronary CT Angiography Using a Deep Learning-Based Technique with Iterative...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cardiology

Background:

• Coronary CT angiography (CCTA) is crucial for diagnosing coronary artery disease.
• Iterative reconstruction techniques reduce noise in CCTA but can be further optimized.
• Deep learning offers potential for advanced image processing in medical imaging.

Purpose of the Study:

• To evaluate the feasibility of a deep learning-based denoising technique for CCTA.
• To assess the combined effect of deep learning and iterative reconstruction on image quality.
• To determine if enhanced image quality impacts diagnostic performance for coronary stenosis.

Main Methods:

• Retrospective analysis of 82 patients undergoing CCTA with iterative reconstruction.
• Development of a U-net based deep learning model (ClariCT.AI) for noise prediction and subtraction.
• Objective (noise, SNR, CNR, ERD) and subjective image quality assessments were performed.

Main Results:

• Denoised CCTA images showed significantly reduced image noise (p < 0.001).
• Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) significantly improved (p < 0.001).
• Subjective image quality scores and edge rise distance (ERD) also showed significant improvements (p < 0.001).

Conclusions:

• Deep learning denoising, combined with iterative reconstruction, effectively enhances CCTA image quality.
• The technique significantly improves objective and subjective image noise reduction and sharpness.
• Diagnostic accuracy for detecting significant coronary stenosis remained comparable.