Computation of the fractional anisotropy and mean diffusivity maps without tensor decoding and diagonalization:...

Area of Science:

• Medical Imaging
• Biophysics
• Neuroscience

Background:

• Diffusion tensor MRI (DT-MRI) enables in vivo mapping of deep tissue organization.
• Commonly used anisotropy maps (mean diffusivity, relative anisotropy, fractional anisotropy) are computationally intensive, typically calculated offline.
• Online computation of these maps would significantly enhance clinical utility for immediate feedback.

Purpose of the Study:

• To derive analytical expressions for computing DT-MRI anisotropy measures online.
• To establish relationships between tensor invariants and moments of diffusion-weighted data.
• To validate these methods using simulations and real-world DT-MRI data.

Main Methods:

• Derived analytical expressions relating anisotropy measures (RA, FA) to the moments of diffusion-weighted (DW) data.
• Utilized the principal icosahedron encoding scheme for deriving relationships.
• Validated findings through Monte Carlo simulations and whole-brain DT-MRI datasets with varying parameters (b-factors, encoding schemes, SNR).

Main Results:

• Established an online computation method for relative anisotropy (RA) based on the mean and standard deviation of DW measurements.
• Derived an analytical expression linking relative anisotropy (RA) and fractional anisotropy (FA).
• Validated the accuracy of the derived methods across diverse experimental conditions, including investigation of encoding scheme biases.

Conclusions:

• Developed a computationally efficient procedure for online calculation of key DT-MRI anisotropy maps (RA, FA).
• The derived analytical expressions facilitate real-time analysis of tissue organization.
• This advancement holds potential for improving the clinical applicability of DT-MRI in time-sensitive scenarios.