Accelerated High Spatial Resolution Diffusion-Weighted Imaging.

Area of Science:

• Medical Imaging
• Biophysics
• Neuroscience

Background:

• Diffusion-weighted imaging (DWI) is crucial for visualizing brain microstructure.
• Current high-resolution DWI methods using anisotropic snapshots face limitations in handling missing data and integrating information across different views.
• Existing reconstruction techniques often process each diffusion-weighted image independently, neglecting anatomical consistency.

Purpose of the Study:

• To develop a novel multi-snapshot DWI reconstruction technique for simultaneous high-resolution reconstruction and local tissue model estimation.
• To enable robust reconstruction from snapshots with varying subsets of diffusion gradients, addressing challenges like intra-scan motion.
• To improve the potential for acceleration in DWI acquisition while maintaining or enhancing image quality.

Main Methods:

• A joint probabilistic model with missing observations was formalized to integrate information from multiple anisotropic snapshots.
• The approach simultaneously reconstructs high-resolution images and estimates tissue models, accounting for missing data.
• The technique was evaluated using synthetic simulations, simulated multi-snapshot scenarios, and in vivo multi-snapshot imaging.

Main Results:

• The combined approach demonstrated superior high-resolution reconstruction compared to conventional methods.
• Tissue model estimation accuracy was significantly improved by the novel reconstruction technique.
• The method allows for quantification of reconstruction errors when snapshots are missing, enhancing reliability.

Conclusions:

• The proposed multi-snapshot DWI reconstruction technique offers improved spatial resolution and tissue characterization.
• The method provides increased robustness to patient motion and potential for faster scanning.
• This advancement holds promise for enhanced in vivo characterization of brain connectivity and microstructure.