PCLR: phase-constrained low-rank model for compressive diffusion-weighted MRI.

Area of Science:

• Magnetic Resonance Imaging
• Biomedical Engineering
• Image Reconstruction

Background:

• Diffusion-weighted (DW) MRI is crucial for neuroimaging.
• Acquisition time and data size are significant limitations in DW MRI.
• Compressive sensing offers a potential solution to accelerate DW MRI acquisition.

Purpose of the Study:

• To develop and validate a novel compressive sensing approach for DW MRI.
• To address challenges like phase inconsistencies and motion artifacts in DW MRI.
• To enable significant reductions in k-space sampling for DW MRI.

Main Methods:

• Developed a phase-constrained low-rank (PCLR) method leveraging image coherence across DW directions.
• Incorporated GRAPPA (generalized autocalibrating partial parallel acquisition) for improved phase estimation and higher undersampling factors.
• Implemented an efficient reconstruction algorithm combining partial Fourier update and singular value decomposition.

Main Results:

• PCLR demonstrated superior performance over frame-independent reconstruction methods.
• Achieved a four-fold undersampling factor with GRAPPA-enhanced phase estimation.
• Error metrics based on diffusion tensor metrics and tractography confirmed PCLR's effectiveness.

Conclusions:

• The PCLR method is effective for compressive DW MRI.
• A four-fold reduction in k-space sampling is achievable with minimal degradation of image quality and diffusion tensor measures.
• This technique facilitates reduced data acquisition and/or improved resolutions in DW MRI.