Improved pixel-by-pixel MRI R2* relaxometry by nonlocal means

Area of Science:

• Medical Imaging
• Biophysics
• Image Processing

Background:

• Quantifying MRI relaxometry pixel-by-pixel is challenging with low signal-to-noise ratio (SNR) images.
• Low-pass Gaussian (LPG) filtering reduces noise but causes blurring.
• Nonlocal means (NLM) algorithm reduces noise while preserving image details.

Purpose of the Study:

• To evaluate the feasibility of enhancing MRI R2* mapping through pre-curve-fitting image filtering.
• To investigate the impact of LPG and NLM filtering on MRI decay signals and R2* mapping.

Main Methods:

• Comparison of LPG and NLM filtering techniques on simulated and in vivo MRI data.
• Pixel-by-pixel curve-fitting was employed for R2* quantification.
• Noise reduction strategies were applied before relaxometry analysis.

Main Results:

• Both LPG and NLM filtering decreased root-mean-square errors in R2* maps.
• NLM filtering preserved map details better than LPG filtering, which introduced blurring.
• NLM consistently produced R2* maps with lower errors than LPG filtering.

Conclusions:

• Pixel-by-pixel fitting can introduce inaccuracies in MRI relaxometry.
• NLM filtering demonstrates superior performance over LPG for MRI R2* mapping.
• NLM holds potential for more accurate pixel-by-pixel MRI relaxometry, enhancing tissue characterization.