Technical note: Revised projections onto convex sets reconstruction of multi-shot diffusion-weighted imaging

Area of Science:

• Medical Imaging
• Image Reconstruction
• Diffusion-Weighted Imaging

Background:

• High-resolution diffusion-weighted imaging (DWI) relies on multi-shot acquisitions and parallel imaging.
• Existing Projections Onto Convex Sets (POCS) algorithms for DWI reconstruction suffer from slow convergence and suboptimal image quality.
• These limitations hinder the widespread application of advanced DWI techniques.

Purpose of the Study:

• To develop a revised Projections Onto Convex Sets (POCS) algorithm for multi-shot DWI reconstruction.
• To enhance convergence speed and accuracy by integrating the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA).
• To improve the practical utility of multi-shot DWI.

Main Methods:

• Modified Projections Onto Convex Sets (POCS) algorithms by incorporating the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA) scheme.
• FISTA improves convergence by utilizing information from two previous iterations.
• The revised approach was applied to POCS-based methods like POCSENSE, POCSMUSE, iPOCSMUSE, and POCS-ICE.
• Performance was evaluated using simulations and in vivo experiments.

Main Results:

• The revised POCS algorithms demonstrated significantly faster convergence compared to original POCS methods.
• In simulations, revised POCS-ICE required approximately 70% fewer iterations to reach a similar normalized root-mean-square error (nRMSE) level.
• The proposed method resulted in improved image quality with reduced artifacts compared to standard POCS.

Conclusions:

• The revised DWI reconstruction methods offer superior convergence rates over traditional POCS-based algorithms.
• Higher image quality is achieved with equivalent iteration counts.
• The proposed FISTA-enhanced POCS approach provides a practical and efficient solution for multi-shot DWI reconstruction.