Accelerated diffusion spectrum imaging in the human brain using compressed sensing.

Area of Science:

• Neuroimaging
• Medical Physics
• Signal Processing

Background:

• Diffusion spectrum imaging (DSI) provides detailed information about white matter microstructure.
• Current DSI acquisition times can be lengthy, limiting clinical applicability.
• Accelerating DSI acquisition without compromising data quality is a significant challenge.

Purpose of the Study:

• To introduce and validate a novel compressed sensing (CS) method for accelerating diffusion spectrum imaging.
• To assess the feasibility of CS for reconstructing sub-Nyquist sampled DSI data in diffusion space.
• To evaluate the impact of CS on preserving essential diffusion properties at accelerated scan rates.

Main Methods:

• Developed a compressed sensing reconstruction framework applicable to diffusion space.
• Simulated 2D and 3D fiber crossings to evaluate CS performance with various undersampling patterns (random, Gaussian, Poisson disk) and acceleration factors (R).
• Acquired DSI data from healthy volunteers, undersampled diffusion space, and reconstructed using CS.

Main Results:

• Compressed sensing successfully reconstructed sub-Nyquist sampled DSI data, preserving essential diffusion properties.
• Key diffusion metrics including orientation distribution function, diffusion coefficient, and kurtosis were maintained up to an acceleration factor of R = 4.
• Different undersampling patterns showed varying effects on diffusion information, highlighting the need for optimized sampling strategies.

Conclusions:

• Compressed sensing offers a viable approach to accelerate diffusion spectrum imaging acquisition.
• The developed method allows for reduced scan times or improved diffusion space resolution without significant loss of critical microstructural information.
• This CS-based acceleration holds promise for enhancing the clinical utility of diffusion spectrum imaging.