Stepwise Stochastic Dictionary Adaptation Improves Microstructure Reconstruction with Orientation Distribution...

Area of Science:

• Neuroimaging
• Biophysics
• Computational Neuroscience

Background:

• Accurate modeling of white matter microstructure is crucial for understanding brain structure and function.
• Multicompartment biophysical models are complex and computationally intensive, often posing ill-posed optimization problems.
• Orientation Distribution Function (ODF) Fingerprinting offers a tractable approach but is limited by dictionary generation methods.

Purpose of the Study:

• To develop an improved method for generating Orientation Distribution Function (ODF) dictionaries for white matter modeling.
• To enhance the computational tractability and accuracy of multicompartment biophysical models in diffusion MRI.
• To specifically tailor ODF dictionary generation to individual diffusion-weighted images.

Main Methods:

• Proposed a stepwise stochastic adaptation mechanism for generating ODF dictionaries.
• Compared the proposed method against traditional uniform distribution sampling for ODF dictionaries.
• Validated the approach using a diffusion phantom and in vivo human brain imaging data.

Main Results:

• The stepwise stochastic adaptation mechanism generated ODF dictionaries tailored to specific diffusion-weighted images.
• Reconstructed diffusivities were significantly less noisy compared to methods using uniform ODF dictionaries.
• Separation of the free water fraction was more pronounced with the proposed dictionary generation method.

Conclusions:

• The proposed stepwise stochastic adaptation mechanism offers a more accurate and robust approach to ODF dictionary generation.
• This method improves the quality of white matter microstructure modeling in diffusion MRI.
• The enhanced separation of free water fraction has implications for more precise brain tissue characterization.