Weighted Schatten p-norm minimization for 3D magnetic resonance images denoising

Area of Science:

• Medical Imaging
• Image Processing
• Computational Science

Background:

• Magnetic Resonance (MR) imaging is crucial for clinical diagnosis and research.
• Rician noise frequently corrupts MR images, hindering accurate analysis.
• Effective noise suppression is vital for reliable MR image interpretation.

Purpose of the Study:

• To develop a novel framework for suppressing Rician noise in 3D MR images.
• To improve the quality of MR images for better clinical and research applications.
• To introduce a low-rank matrix approximation (LRMA) method with weighted Schatten p-norm minimization (WSNMD-3D).

Main Methods:

• A framework utilizing low-rank matrix approximation (LRMA) with weighted Schatten p-norm minimization (WSNMD-3D) was proposed.
• Similar non-local cubic patches from noisy 3D MR images were grouped into a matrix.
• Weighted Schatten p-norm minimization (WSNM) was applied to approximate the latent low-rank matrix, considering different rank components.

Main Results:

• The proposed WSNMD-3D method demonstrated superior performance in denoising 3D MR images compared to existing state-of-the-art techniques.
• Both visual inspection and quantitative evaluations confirmed the effectiveness of the denoising approach.
• The method successfully approximated the true rank of the latent low-rank matrix.

Conclusions:

• The WSNMD-3D framework offers an effective solution for Rician noise reduction in 3D MR images.
• The approach enhances image quality, aiding clinical diagnosis and scientific research.
• This method provides a significant advancement in MR image denoising techniques.