Simple noise reduction for diffusion weighted images.

Area of Science:

• Medical Imaging
• Biophysics

Background:

• Diffusion-weighted (Dw) imaging is crucial for visualizing water molecule diffusion.
• High b-value Dw images often suffer from noise, reducing signal-to-noise ratio (SNR).
• Accurate noise reduction is essential for reliable Dw imaging analysis.

Purpose of the Study:

• To reduce noise and enhance SNR in high b-value Dw images.
• To compare the efficacy of Rician and Gaussian distribution-based noise correction schemes.
• To determine the optimal noise correction method for Dw imaging.

Main Methods:

• Utilized phantoms with varying sucrose concentrations (10, 30, 50 wt%).
• Implemented and compared two noise correction schemes: Rician and Gaussian distribution-based.
• Quantified errors in noise reduction across different concentrations for both schemes.

Main Results:

• The Rician distribution scheme yielded significantly lower highest error values: 7.3% (10 wt%), 2.4% (30 wt%), and 0.1% (50 wt%).
• The Gaussian distribution scheme resulted in higher errors: 20.3% (10 wt%), 11.6% (30 wt%), and 3.4% (50 wt%).
• The Rician scheme demonstrated superior performance in noise reduction for Dw imaging.

Conclusions:

• The Rician distribution-based noise correction scheme is highly effective for improving SNR in high b-value Dw imaging.
• Noise reduction using the Rician scheme enables more accurate application of Dw imaging corrections.
• This study highlights the importance of selecting appropriate noise models for quantitative Dw imaging.