[Brain tissue microstructure parameters estimation method based on proximal gradient network].

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Computational Neuroscience

Background:

• Diffusion tensor imaging (DTI) offers insights into white matter structure but lacks detailed brain tissue microstructure information.
• Neurite orientation dispersion and density imaging (NODDI) enhances microstructure details but requires numerous diffusion gradients and complex calculations.
• Existing methods like maximum likelihood fitting are computationally intensive and time-consuming for accurate microstructure parameter estimation.

Purpose of the Study:

• To develop an advanced method for estimating brain tissue microstructure parameters.
• To overcome the limitations of existing imaging techniques regarding data acquisition and computational complexity.
• To improve the accuracy and efficiency of brain microstructure imaging.

Main Methods:

• Proposed a novel microstructure parameter estimation method utilizing a proximal gradient network.
• This approach avoids traditional fitting paradigms, reducing reliance on extensive diffusion gradients.
• The network is trained to directly estimate microstructure parameters from diffusion imaging data.

Main Results:

• The proximal gradient network accurately estimates microstructure parameters.
• The method significantly reduces the number of required diffusion gradients compared to conventional techniques.
• Achieved superior imaging quality compared to neurite orientation dispersion and density imaging (NODDI) and accelerated microstructure imaging via convex optimization (aMICON) models.

Conclusions:

• The proposed proximal gradient network offers an efficient and accurate solution for brain microstructure imaging.
• This technique enhances imaging quality while reducing acquisition time and computational load.
• Represents a significant advancement over existing DTI and NODDI-based microstructure analysis methods.