MR Image Reconstruction Using a Combination of Compressed Sensing and Partial Fourier Acquisition: ESPReSSo.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Compressed Sensing

Background:

• Accelerated MRI acquisition is crucial for reducing scan times.
• Standard subsampling techniques can lead to aliasing artifacts and reduced image quality, particularly in high-frequency components.

Purpose of the Study:

• To introduce and evaluate a new Cartesian subsampling scheme, ESPReSSo, for improved MRI reconstruction.
• To enhance k-space sampling density and improve image quality, especially edge delineation.

Main Methods:

• ESPReSSo combines partial Fourier (PF) acquisition with variable-density Poisson Disc (vdPD) subsampling.
• Reconstruction utilizes a Compressed Sensing (CS) algorithm with a Hermitian symmetry constraint.
• The method was validated on in-vivo abdominal MRI datasets with various reconstruction strategies.

Main Results:

• ESPReSSo effectively redistributes k-space sampling, increasing density in critical regions.
• Improved edge delineation and regional homogeneity were observed in reconstructed images.
• Analysis of global and local image metrics confirmed the clinical feasibility of the setup.

Conclusions:

• ESPReSSo offers a clinically feasible approach for accelerated MRI acquisition.
• The technique is beneficial for applications requiring precise tissue boundaries, such as segmentation and lesion detection.