Detection and modeling of non-Gaussian apparent diffusion coefficient profiles in human brain data

Area of Science:

• Magnetic Resonance Imaging
• Diffusion MRI
• Computational Neuroscience

Background:

• Standard diffusion tensor imaging (DTI) assumes Gaussian diffusion, which is insufficient for complex biological tissues.
• Non-Gaussian diffusion effects occur at microstructural interfaces, such as white matter tract crossings.
• Accurate modeling of diffusion is crucial for understanding tissue microstructure and detecting abnormalities.

Purpose of the Study:

• To develop and validate a method for modeling non-Gaussian apparent diffusion coefficient (ADC) profiles in diffusion-weighted MRI.
• To classify voxel-wise diffusion profiles as isotropic, anisotropic Gaussian, or non-Gaussian.
• To demonstrate the utility of this method in identifying complex tissue structures and potential pathologies in the human brain.

Main Methods:

• Acquisition of multi-directional diffusion-weighted MR data with standard imaging parameters (b ≈ 1000 s/mm²).
• Modeling of ADC profiles using spherical harmonic (SH) expansion with varying orders.
• Development of a model selection strategy to balance data adequacy and prevent overfitting.
• Classification of diffusion profiles based on the underlying probability density function of water molecule displacement.

Main Results:

• Non-Gaussian ADC profiles were consistently observed in brain regions with known complex tissue structures.
• The developed method successfully classified diffusion profiles in data acquired on clinical scanners.
• Synthetic data validation confirmed the genuine nature of the observed non-Gaussian effects.
• Model complexity was found to decrease in the presence of simulated pathology.

Conclusions:

• The proposed method effectively captures non-Gaussian diffusion effects, surpassing the limitations of the standard diffusion tensor model.
• This technique provides a robust classification of tissue microstructure based on diffusion MRI data.
• The method shows promise as a sensitive indicator of structural changes and pathology affecting tissue complexity in the brain.