Related Experiment Video
Updated: Oct 26, 2026
Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging
Published on: November 8, 2012
Denoising of multi b-value diffusion-weighted MR imaging using an improved non-local means filter
Yu-Chun Lin1, Chi-Kuang Liu2, Hsuan-Ming Huang1
1Institute of Medical Device and Imaging, College of Medicine, National Taiwan University, No.1, Sec. 1, Jen Ai Rd., Zhongzheng Dist., Taipei City 100, Taiwan; Program for Precision Health and Intelligent Medicine, Graduate School of Advanced Technology, National Taiwan University, MK Innovation Hall, No.1, Sec. 4, Roosevelt Rd., Taipei City 106319, Taiwan.
Purpose:
Intravoxel incoherent motion (IVIM) maps derived from the quantitative analysis of multiple b-value diffusion-weighted magnetic resonance imaging (DW-MRI) has shown many clinical applications. However, DW-MRI is affected by Rician noise, which compromises the reliability of IVIM maps.
Method:
To address this issue, we propose an improved non-local means (NLM) filtering method that utilizes a prior image to improve the estimation of similarity weights. Moreover, the proposed method is implemented using a fast version of NLM filtering to efficiently perform image denoising. We used both simulated and real DW-MRI data to evaluate the performance of the proposed method and compared the results with those obtained using three image denoising methods: local principal component analysis, Marchenko-Pastur principal component analysis, and pre-filtered rotationally invariant non-local principal component analysis.
Results:
Results from the simulated DW-MRI data showed that the proposed method outperformed the other three methods in terms of root mean-squared error and improved the quality of IVIM maps. Results from the real DW-MRI data showed that the proposed method yielded IVIM parameter values consistent with those reported in the literature and exhibited lower coefficients of variation compared to the other three methods.
Conclusion:
Overall, the proposed NLM filtering method has the potential to be a fast and reliable denoising technique for multiple b-value DW-MRI.
