Accelerating compressed sensing reconstruction of subsampled radial k-space data using geometrically-derived density...

Area of Science:

• Medical Imaging
• Image Reconstruction
• Compressed Sensing

Background:

• Compressed sensing (CS) reconstruction of subsampled radial k-space data is crucial for accelerated imaging.
• Existing density compensation functions (DCFs) may not optimally balance reconstruction speed and image quality.

Purpose of the Study:

• To develop and evaluate a novel geometrically-derived density compensation function (gDCF) for accelerating CS reconstruction.
• To assess the impact of gDCF on image quality and reconstruction time compared to standard methods.

Main Methods:

• A theoretical framework was established to compute gDCF based on Nyquist distance in polar coordinates.
• gDCF performance was evaluated against standard DCF and modified Shepp-Logan filters using phantom and cardiac MRI datasets.
• Image quality was quantified using NRMSE, blur metrics, and SSIM; reconstruction time was also measured.

Main Results:

• Phantom data showed good image quality metrics for all tested DCFs, validating gDCF's uniform density at Nyquist.
• CS reconstruction with gDCF yielded significantly higher SSIM and NRMSE compared to other DCFs for cardiac MRI.
• gDCF significantly reduced reconstruction time compared to no DCF, standard DCF, and modified SL filter.

Conclusions:

• The proposed gDCF effectively accelerates CS reconstruction of subsampled radial k-space data.
• gDCF achieves this acceleration without compromising image quality, offering a significant improvement over existing methods.