Shearlet-based compressed sensing for fast 3D cardiac MR imaging using iterative reweighting.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiology

Background:

• High-resolution 3D cardiovascular magnetic resonance (CMR) imaging is crucial but limited by long acquisition times.
• Compressed sensing (CS) offers a potential solution to accelerate CMR acquisition.
• Developing efficient sparsifying transforms is key for successful CS reconstruction in CMR.

Purpose of the Study:

• To introduce and evaluate a novel CS reconstruction approach, 3DShearCS, for accelerated 3D CMR.
• To assess the performance of shearlets as a sparsifying transform in 3D CMR.
• To compare 3DShearCS with existing reconstruction methods for speed and image quality.

Main Methods:

• Developed a novel compressed sensing (CS) reconstruction algorithm named 3DShearCS, utilizing shearlets as the sparsifying transform.
• Employed 3D radial phase encoding (RPE) for data acquisition.
• Implemented an iterative reweighting scheme for enhanced convergence and image quality during reconstruction.
• Compared 3DShearCS against three other common reconstruction approaches.

Main Results:

• 3DShearCS demonstrated lower relative errors and higher structural similarity compared to other methods.
• The technique achieved higher diagnostic scores from expert reviewers, particularly at high undersampling factors (short scan times).
• Shearlets proved more efficient than traditional wavelets for this application.

Conclusions:

• 3DShearCS enables fast, high-resolution 3D CMR acquisition without compromising anatomical depiction.
• The proposed method effectively accelerates CMR imaging, potentially increasing its clinical adoption.
• Shearlet-based CS reconstruction is a promising technique for overcoming acquisition time limitations in 3D CMR.