HICU is a computational procedure for accelerated, calibrationless magnetic resonance image reconstruction that is fast, memory efficient, and ready to scale to highdimensional imaging. For demonstration, HICU is applied to multi-coil 2D static imaging, multi-coil 2D dyanmic (2D+T) imaging, and 3D knee imaging.
- Shen Zhao, The Ohio State University (zhao.1758@osu.edu)
- Lee C. Potter, The Ohio State University (potter.36@osu.edu)
- Rizwan Ahmad, The Ohio State University (ahmad.46@osu.edu)
- One 2D T2-weighted parallel MRI k-space dataset (https://fastmri.med.nyu.edu/) and four sampling masks are in the folder 2D/2D_data.
- The HICU reconstruction subroutine for 2D is in 2D/HICUsubroutine_2D.m.
- The optional nondecimated wavelet denoiser is in the file 2D/SWT_denoiser.m.
- To implement the reconstruction for 2D, run file 2D/main.m.
- One 2D+T cardiac cine parallel MRI k-space and two sampling masks are in 2D_T/2D+T_Data.
- The HICU reconstruction subroutine for 2D+T is in 2D+T/HICUsubroutine_2D_T.m.
- To implement the reconstruction for 2D+T, run file 2D+T/main.m
- One 3D knee parallel MRI k-space and two sampling masks are in 3D/3D_Data.
- The HICU reconstruction subroutine for 3D is in 3D/HICUsubroutine_3D.m.
- To implement the reconstruction for 3D, run file 3D/main.m
- https://arxiv.org/abs/2002.03225
- https://arxiv.org/abs/2004.08962
- https://ieeexplore.ieee.org/document/9433815
- https://onlinelibrary.wiley.com/doi/10.1002/mrm.28721 (The Algorithm 1 is missing in this link but can be found in the arXiv version.)