Nonlocal maximum likelihood estimation method for denoising multiple-coil magnetic resonance images.

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Effective denoising is crucial for accurate analysis and quantitative measurements in magnetic resonance (MR) imaging.
• Existing denoising methods often fail to address MR images acquired with phased-array coils, especially when k-space is subsampled.
• Subsampling in k-space, common in accelerated acquisition techniques like GRAPPA, introduces spatially varying noise.

Purpose of the Study:

• To propose and validate a novel method for denoising MR images acquired using multiple phased-array coils.
• To account for both the non-central chi distribution of magnitude data and the spatially varying noise characteristics.

Main Methods:

• Developed a denoising method specifically designed for multiple-coil MR image acquisition.
• Incorporated the non-central chi distribution model for magnitude data.
• Addressed the challenge of spatially varying noise introduced by k-space subsampling.

Main Results:

• The proposed method effectively denoises MR images acquired with phased-array coils.
• Experimental results on simulated and real data demonstrate the method's superior performance.
• Accurate estimation of true signal from noisy MR images is achieved.

Conclusions:

• The developed denoising technique offers a significant advancement for MR image processing.
• It provides a robust solution for analyzing images acquired with accelerated techniques using phased-array coils.
• This method enhances the reliability of quantitative measurements derived from MR images.