Related Experiment Video
Updated: Apr 16, 2026
09:33
Diffusion Tensor Magnetic Resonance Imaging in the Analysis of Neurodegenerative Diseases
Published on: July 28, 2013
29.5K
A fast algorithm for denoising magnitude diffusion-weighted images with rank and edge constraints
Fan Lam1,2, Ding Liu1,2, Zhuang Song3
1Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA.
Magnetic Resonance in Medicine
|March 4, 2015
Summary
A new majorize-minimize algorithm significantly accelerates denoising of diffusion-weighted images by a factor of 10. This method efficiently handles joint rank and edge constraints, improving computational speed for large datasets.
Area of Science:
- Medical Imaging
- Computational Neuroscience
- Image Processing
Background:
- Diffusion-weighted imaging (DWI) is crucial for neuroimaging.
- Denoising DWI is essential for accurate analysis.
- Existing methods struggle with computational efficiency for complex constraints.
Purpose of the Study:
- To develop an accelerated denoising algorithm for magnitude diffusion-weighted images.
- To incorporate joint rank and edge constraints into the denoising process.
- To improve the computational efficiency of DWI denoising.
Main Methods:
- Extension of a majorize-minimize method for statistical estimation.
- Incorporation of joint rank and edge constraints using noncentral chi distributions.
- Decomposition of the problem into a series of constrained Gaussian denoising problems solved via alternating minimization.
Main Results:
- Achieved a 10-fold speed-up compared to the original Quasi-Newton-based algorithm.
- Enabled efficient denoising of large DWI datasets with multiple diffusion directions.
- Demonstrated comparable or superior denoising performance to the original algorithm.
- Showcased denoising effects on fiber tracking results around the hippocampus in vivo.
Conclusions:
- The proposed majorize-minimize algorithm efficiently solves the optimization problem for denoising DWI.
- The algorithm effectively handles noncentral chi distributions with joint rank and edge constraints.
- This advancement offers a computationally efficient solution for complex DWI denoising tasks.
