Keywords: Synthetic MR, AI/ML Image Reconstruction, Generative Models, Synthetic MR
Motivation: Deep learning based MRI reconstruction methods require access to large amounts of high quality paired training data including raw k-space.
Goal(s): Present a method to generate multiple synthetic raw datasets from a single magnitude image.
Approach: A generative model is developed/trained on FSE/GRE, 2D/3D acquired data to generate multiple plausible synthetic phase images from one magnitude image. These data are then combined with randomly chosen sensitivity maps from an RF coil library to create synthetic raw k-space.
Results: Synthetic and ground truth phase images are close in distribution; when combined to make raw k-space are performant as training data for downstream image reconstruction.
Impact: The ability to create training datasets from clinical archives to train custom downstream reconstruction models without reliance on prospectively made datasets for training
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