5D image reconstruction exploiting space-motion-echo sparsity for accelerated free-breathing quantitative liver MRI

Area of Science:

• Magnetic Resonance Imaging
• Medical Physics
• Image Reconstruction

Background:

• Free-breathing 3D non-Cartesian multi-echo gradient-echo (mGRE) imaging enables liver tissue MR parameter mapping.
• Compressed sensing (CS)-based 4D motion-resolved reconstruction allows high-resolution proton density fat fraction (PDFF), R2*, and quantitative susceptibility mapping (QSM).
• Long scan times limit the clinical utility of current free-breathing 3D non-Cartesian mGRE techniques.

Purpose of the Study:

• To develop and evaluate a CS-based 5D motion-resolved mGRE image reconstruction method to accelerate free-breathing liver imaging.
• To investigate the potential of leveraging 5D data correlations for improved MR parameter mapping.
• To compare the performance of 5D reconstruction against 4D reconstruction across various acceleration factors.

Main Methods:

• Proposed a CS-based 5D motion-resolved mGRE reconstruction integrating discrete wavelet transforms.
• Developed a solution algorithm for processing 5D complex-valued arrays.
• Evaluated the method using phantom and in vivo human subject studies, comparing 5D and 4D reconstructions.

Main Results:

• The 5D reconstruction demonstrated improved reliability and consistency in PDFF, R2*, and QSM measurements compared to 4D reconstruction.
• The proposed method effectively leverages unexplored correlations within the 5D data.
• More consistent results were observed across a range of acceleration factors with 5D reconstruction.

Conclusions:

• The 5D motion-resolved image reconstruction is feasible for accelerated and reliable free-breathing liver mGRE imaging.
• This technique enhances the measurement accuracy of PDFF, R2*, and QSM.
• The 5D approach offers a promising advancement for quantitative liver MRI.