Self Supervised Denoising Diffusion Probabilistic Models for Abdominal DW-MRI.

Area of Science:

• Medical Imaging
• Radiology
• Biomedical Engineering

Background:

• Quantitative diffusion-weighted MRI (DW-MRI) is crucial for abdominal disease detection.
• Low signal-to-noise ratio (SNR) at high b-values limits DW-MRI accuracy.
• Current methods to improve SNR, like averaging multiple directions, increase scan time and introduce artifacts.

Purpose of the Study:

• To develop a novel parameter estimation technique for denoising diffusion-weighted images (DWIs).
• To enable accurate DW-MRI with single diffusion gradient direction images.
• To overcome SNR limitations in abdominal quantitative diffusion MRI.

Main Methods:

• Proposed a self-supervised diffusion denoising probabilistic model (ssDDPM).
• The model effectively denoises diffusion-weighted images.
• The technique works on single diffusion gradient direction images, reducing scan time.

Main Results:

• The ssDDPM effectively denoises abdominal DWIs.
• Improved SNR and image quality were achieved.
• Accurate parameter estimation is possible even with single-direction images.

Conclusions:

• The proposed ssDDPM offers a promising solution for accurate abdominal quantitative diffusion MRI.
• This method addresses SNR limitations and reduces artifacts.
• The technique facilitates faster and more reliable disease assessment using DW-MRI.