Correcting motion artifacts in coronary computed tomography angiography images using a dual-zone cycle generative...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Cardiovascular Diagnostics

Background:

• Coronary computed tomography angiography (CCTA) is crucial for diagnosing coronary artery disease.
• Motion artifacts in CCTA images, particularly from 64-slice spiral CT, can hinder accurate diagnosis.
• These artifacts often affect critical coronary artery segments like the right coronary artery and left circumflex artery.

Purpose of the Study:

• To develop and evaluate a novel method for correcting motion artifacts in clinical CCTA images.
• The objective is to enhance the diagnostic quality of CCTA scans affected by motion.

Main Methods:

• A generative adversarial network (GAN), specifically a dual-zone GAN, was proposed for artifact correction.
• The GAN was trained using pairs of CCTA regions of interest (ROIs) or full images with and without motion artifacts.
• The trained network corrects artifacts by inputting only the motion-affected clinical CCTA images.

Main Results:

• The dual-zone GAN significantly improved image quality metrics, including peak signal to noise ratio (PSNR) and structural similarity (SSIM), while reducing mean square error (MSE) and mean absolute error (MAE).
• Quantitative improvements were observed for both ROIs and full images compared to the original data.
• Physician evaluations indicated higher scores for the artifact correction quality of the generated images.

Conclusions:

• The dual-zone GAN demonstrates a strong capability for correcting motion artifacts in coronary arteries within CCTA images.
• The method effectively preserves the essential characteristics of the original clinical CCTA images.
• This approach holds promise for improving the reliability of CCTA in diagnosing coronary artery disease.