Enhanced spectral response in frequency-dependent diffusion measurements using a linear encoding model

Area of Science:

• Magnetic Resonance Imaging
• Diffusion MRI
• Quantitative Imaging

Background:

• Diffusion MRI measures water diffusion in biological tissues.
• Frequency-dependent diffusion (D(ω)) provides insights into tissue microstructure.
• Current methods often use simplified single-frequency representations for spectral encodings.

Purpose of the Study:

• To develop a more comprehensive quantitative representation for spectral encodings in diffusion MRI.
• To improve the estimation accuracy of frequency-dependent diffusion (D(ω)).

Main Methods:

• Representing spectral response using encoding power over contiguous frequency intervals (|Q(ω)|²).
• Formulating a linear encoding model for estimating interval-specific diffusivity.
• Validating the strategy in vivo using human brain imaging and simulations.
• Utilizing a framework for selecting robust spectral encodings.

Main Results:

• In vivo D(ω) increased with frequency and showed more spectral selectivity compared to single-frequency methods.
• Simulations indicated improved D(ω) estimation accuracy, particularly for nonlinear D(ω).
• Reduced precision was observed for narrower frequency intervals.

Conclusions:

• A holistic representation paradigm for spectral diffusion measurements enhances D(ω) recovery fidelity.
• The proposed method offers a more accurate approach to characterizing frequency-dependent diffusion.
• This advancement has implications for understanding tissue microstructure in diffusion MRI.