A simple application of compressed sensing to further accelerate partially parallel imaging

Area of Science:

• Magnetic Resonance Imaging
• Image Reconstruction
• Signal Processing

Background:

• Compressed sensing (CS) and partially parallel imaging (PPI) accelerate magnetic resonance (MR) imaging by reducing k-space data acquisition.
• Combining CS and PPI has been challenging due to CS's incoherent sampling requirement and PPI's regular sampling grids.

Purpose of the Study:

• To develop and validate a novel method, "CS+GRAPPA," that integrates CS with GRAPPA for enhanced MR imaging.
• To overcome the limitations of combining CS and PPI on regularly sampled k-space data.

Main Methods:

• Decomposition of equidistant k-space samples into random subsets for CS reconstruction.
• Application of standard CS and edge- and joint-sparsity-guided CS reconstructions.
• Calibration of GRAPPA coil weights using CS reconstructions and evaluation using the Case-PDM metric on in vivo data.

Main Results:

• CS+GRAPPA significantly reduced coherent aliasing and noise artifacts compared to standard GRAPPA.
• Edge- and joint-sparsity-guided CS minimized blurring introduced by increased subset decomposition.
• The proposed method achieved similar image quality to standard GRAPPA with approximately half the number of samples.

Conclusions:

• CS+GRAPPA effectively combines CS and GRAPPA, enabling CS reconstruction even with equidistant k-space sampling.
• The novel method offers a significant improvement in MR image quality and acceleration potential.
• This technique holds promise for faster and more robust MR imaging applications.