Res-MoCoDiff: Residual-guided diffusion models for motion artifact correction in brain MRI.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Science

Background:

• Motion artifacts degrade MRI quality, impacting diagnostic accuracy.
• Current correction methods are time-consuming and costly.
• Developing efficient artifact correction is crucial for clinical MRI.

Purpose of the Study:

• Introduce Res-MoCoDiff, an efficient denoising diffusion probabilistic model for MRI motion artifact correction.
• Address limitations of conventional MRI artifact mitigation techniques.
• Enhance the utility of MRI for quantitative analysis and clinical applications.

Main Methods:

• Developed Res-MoCoDiff, a diffusion model with a U-net backbone and Swin Transformer blocks.
• Implemented a novel residual error shifting mechanism for noise simulation.
• Utilized a combined L1 + L2 loss function for improved image sharpness.
• Evaluated on in-silico and in-vivo datasets, comparing against CycleGAN, Pix2pix, and MT-DDPM.

Main Results:

• Res-MoCoDiff demonstrated superior artifact removal across all distortion levels.
• Achieved highest SSIM and lowest NMSE, with PSNR up to 41.91 dB.
• Reduced average sampling time to 0.37 seconds per batch, a significant improvement over conventional methods.

Conclusions:

• Res-MoCoDiff provides a robust and efficient solution for MRI motion artifact correction.
• The model preserves fine structural details and reduces computational overhead.
• Potential for seamless clinical integration to improve diagnostic accuracy and patient care.