Keywords: Image Reconstruction, AI/ML Image Reconstruction
Motivation: Deep learning-based MRI reconstruction often requires extensive training data or fully sampled ground truths, potentially limiting flexibility and applicability.
Goal(s): To develop a novel zero-shot learning method for undersampled MRI reconstruction without the need for training data.
Approach: The method trains a denoiser by randomly splitting acquired undersampled k-space into two complementary subsets, applying unbiased reconstruction to each subset to create training data and target labels, respectively, and performing Noise2Noise training. The trained denoiser is then integrated into the plug-and-play framework for final reconstruction.
Results: Experiments demonstrate advantages over existing techniques, offering improved flexibility and adaptability, particularly in scenarios with limited training data.
Impact: The proposed zero-shot method offers improved MRI reconstruction without requiring any training data. This framework not only provides enhanced image quality and adaptability across diverse MR applications but also is able to extend to other imaging tasks beyond MRI.
How to access this content:
For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.
After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.
After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.
Keywords