Combining variational mode decomposition with regularisation techniques to denoise MRI data.

Area of Science:

• Medical Imaging
• Signal Processing
• Computational Science

Background:

• MRI data often suffers from noise, which can degrade image quality and diagnostic accuracy.
• Traditional denoising methods may struggle with complex noise patterns, particularly Rician noise inherent in MRI.
• Regularization techniques are crucial for solving ill-posed problems like image denoising.

Purpose of the Study:

• To propose a novel and effective method for denoising Magnetic Resonance Imaging (MRI) data.
• To enhance the quality of MRI images by suppressing Rician noise.
• To evaluate the performance of the proposed denoising method against existing state-of-the-art techniques.

Main Methods:

• The proposed method combines regularization techniques with variational mode decomposition (VMD).
• A 2D VMD algorithm is employed to decompose the MR imaging data into intrinsic modes.
• The fused lasso signal approximator is applied to the decomposed modes for noise suppression.

Main Results:

• The novel method demonstrated superior performance in denoising MRI data compared to reference methods.
• Quantitative metrics including PSNR, SSIM, HEN, QILV, and sharpness index confirmed the effectiveness.
• The algorithm showed particularly high denoising performance under heavy noise conditions, validated on both simulated and real MRI images.

Conclusions:

• The proposed regularization and VMD-based approach offers a powerful solution for MRI denoising.
• This method significantly improves the signal-to-noise ratio and structural integrity of MRI images.
• The findings suggest this technique is highly valuable for enhancing medical image analysis and interpretation.