Deep subspace unrolling network for accelerating non-Cartesian sampled CMR Multitasking imaging
Motivation: The prolonged imaging time of CMR Multitasking limits its clinical application. Goal(s): Developing deep learning method to improve the imaging speed while ensuring the reconstruction and quantification accuracy. Approach: Developing a deep subspace unrolling network with an Unet as density compensation to accelerate convergence. Results: The proposed unrolling network achieved the iterative reconstruction in 5 iterations. Impact: The proposed method reduces the imaging time of CMR multitasking from 71 minutes (MATLAB) to 5 minutes (GPU), providing possible clinical applications.
Authors
- Sen Jia (ORCID: https://orcid.org/0000-0002-0953-4960)
- Jing Cheng (ORCID: https://orcid.org/0000-0001-9098-8048)
- Junpu Hu
- Ye Li (ORCID: https://orcid.org/0000-0002-2898-5345)
- Dong Liang (ORCID: https://orcid.org/0000-0002-6712-6631)
- Yining Wang (ORCID: https://orcid.org/0000-0001-6441-2002)
- Qi Liu (ORCID: https://orcid.org/0000-0003-0110-7750)
- Jian Xu (ORCID: https://orcid.org/0000-0002-0307-3198)
- Jiaying Zhao (ORCID: https://orcid.org/0000-0002-3375-4621)
Publication Details
- Journal
- Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition
- Published
- 2025-09-16
- DOI
- https://doi.org/10.58530/2025/2594
- Primary Topic
- Medical Imaging Techniques and Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00