Diffusion tensor imaging using partial Fourier STEAM MRI with projection onto convex subsets reconstruction.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Neuroimaging
• Diffusion Tensor Imaging (DTI)

Background:

• Diffusion-weighted single-shot STEAM MRI offers resonance offset insensitivity for brain diffusion mapping.
• However, its lower signal-to-noise ratio (SNR) and speed compared to echo-planar imaging (EPI) limit its clinical utility.

Purpose of the Study:

• To improve the SNR and speed of diffusion-weighted single-shot STEAM MRI.
• To overcome phase distortions in partial Fourier diffusion-weighted acquisitions.
• To enable robust diffusion mapping and fiber tracking in the human brain.

Main Methods:

• Implementation of partial Fourier encoding combined with projection onto convex subsets (PRO-Sets) reconstruction.
• Optimization of variable flip angles for STEAM readout pulses.
• Acquisition of diffusion-weighted images with optimized parameters for anisotropy studies and fiber tracking.

Main Results:

• A 20% gain in relative SNR per unit time was achieved with 5/8 phase encoding and optimized variable flip angles.
• Imaging time was reduced from 670 ms to 440 ms.
• Demonstrated feasibility for clinical anisotropy studies and fiber tracking with reduced scan times.

Conclusions:

• The combined use of partial Fourier encoding and PRO-Sets effectively compensates for lower SNR and speed in STEAM MRI.
• This technique overcomes phase distortions, enabling accurate diffusion mapping.
• The optimized protocols facilitate efficient clinical applications in neuroimaging.