Denoising of complex MRI data by wavelet-domain filtering: application to high-b-value diffusion-weighted imaging

Area of Science:

• Medical Imaging
• Signal Processing

Background:

• Rician noise in magnitude magnetic resonance (MR) images complicates analysis, especially in low signal-to-noise ratio (SNR) areas.
• This noise creates a false minimum signal, obscuring true low signals and hindering accurate quantification.

Purpose of the Study:

• To develop and evaluate a novel noise reduction technique for MR images.
• To address the challenges posed by Rician noise in low SNR regions.

Main Methods:

• Applied Wiener-like filtering in the wavelet domain to complex MR data prior to magnitude image reconstruction.
• Tested the noise-reduction algorithm on simulated and experimental diffusion-weighted (DW) MR images.

Main Results:

• Significantly reduced signal standard deviation (up to 87% in simulated images).
• Decreased the background noise floor by approximately a factor of 6 in both simulated and experimental images.

Conclusions:

• Wavelet-domain filtering effectively reduces Rician noise in MR images.
• The proposed denoising method enhances image quality and aids in the quantification of low SNR regions.