Keywords: Machine Learning/Artificial Intelligence, AI/ML Image Reconstruction
Motivation: Current deep-learning and prior-image-based four-dimensional MRI (4D-MRI) reconstruction methods lack generalizability, limiting its clinical adoption.
Goal(s): To develop a generalizable deep-learning-based 4D-MRI reconstruction workflow.
Approach: We propose RAUQ-4DRecon, a workflow trained on XCAT-generated data with diverse contrasts and physiological parameters. Structural innovations include a segmentation-guided vertebra affine alignment network and a deformable motion estimation hypernetwork to capture respiratory motion of commercially acquired 4D-MRI and transfer it to pre-acquired ultra-quality 3D-MRI.
Results: RAUQ-4DRecon surpassed VoxelMorph and pTV on all XCAT validation metrics. In real-patient external validation, RAUQ-4DRecon achieved NMI of 0.382±0.071 and vertebra stability (LCC) of 0.347±0.061, compared to DDEM’s 0.328±0.085 and 0.275±0.082, respectively.
Impact: This study addresses data limitations in training 4D-MRI reconstruction workflows, significantly enhancing generalizability of 4D-MRI and expanding its potential clinical applications across various medical settings. The digital phantom-based training procedure also offers valuable insights for other medical image processing tasks.
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