Denoising Diffusion MRI via Graph Total Variance in Spatioangular Domain.

Area of Science:

• Medical Imaging
• Neuroscience
• Signal Processing

Background:

• Diffusion MRI (DMRI) is crucial for diagnosing white matter disorders.
• DMRI data is often corrupted by significant noise, limiting quantitative analysis.
• Current denoising methods using total variance (TV) primarily focus on the spatial domain, neglecting the spatioangular nature of DMRI data.

Purpose of the Study:

• To introduce a novel noise reduction technique for DMRI data.
• To address the limitations of existing spatial-domain denoising methods.
• To improve the quantitative analysis of DMRI by enhancing noise reduction.

Main Methods:

• Representing DMRI data as a graph in the spatioangular domain.
• Applying graph total variance (GTV) regularization to penalize noise-induced variances on the graph.
• Developing a denoising method that leverages both spatial and angular information.

Main Results:

• The proposed GTV method effectively reduces noise in DMRI data.
• GTV outperforms existing state-of-the-art denoising methods.
• Experiments on synthetic and real DMRI data validate the effectiveness of GTV.

Conclusions:

• Graph total variance (GTV) offers a superior approach to denoising DMRI data.
• Incorporating spatioangular information significantly enhances noise reduction efficacy.
• The GTV method holds promise for improving the diagnosis of brain disorders using DMRI.