Improved compressed sensing reconstruction for F magnetic resonance imaging

Area of Science:

• Medical Imaging
• Signal Processing
• Biophysics

Background:

• Compressed sensing (CS) in magnetic resonance imaging (MRI) allows for faster data acquisition by reconstructing undersampled k-space data.
• ¹⁹F MRI data is spatially sparse but often exhibits low signal-to-noise ratio (SNR), leading to artifacts in CS reconstructions and reduced image quality.

Purpose of the Study:

• To develop and evaluate methods for improving the image quality of undersampled, reconstructed CS data in ¹⁹F MRI.
• To address artifacts caused by low SNR in CS reconstructions.

Main Methods:

• Two resampling strategies were developed and combined with CS reconstructions.
• Numerical simulations were conducted using low-SNR, spatially sparse ¹⁹F chemical-shift imaging data.
• Various undersampling factors and SNR levels were tested, with error quantified using root-mean-square error.

Main Results:

• Both proposed resampling strategies demonstrated improved overall image quality compared to conventional CS reconstructions.
• Significant suppression of spike artifacts in the background was observed.
• Minimal changes in signal pixels were maintained.

Conclusions:

• The presented resampling methods effectively enhance the quality of CS reconstructions in ¹⁹F MRI.
• These post-processing techniques require no additional measurement time, facilitating integration into existing MRI protocols and application to previously acquired data.