Keywords: Analysis/Processing, Machine Learning/Artificial Intelligence
Motivation: The FORCE registry of single-ventricle patients contains >5,000 MRI scans, requiring standardized segmentation of five major blood vessels for flow quantification to predict patient outcomes—a challenging task due to complex single-ventricle anatomy.
Goal(s): Automate flow quantification for the left and right pulmonary arteries, aorta, and superior and inferior vena cavae.
Approach: Develop a unified deep learning model and automated pipeline that extracts phase-contrast flow planes from the registry and performs simultaneous classification and segmentation of the five vessels.
Results: MultiFlowSeg achieved a median Dice score of 0.91 on 50 test sets and 89% segmentation success across 630 exams processed through the pipeline.
Impact: We have developed a pipeline that uses our novel MultiFlowSeg model to automate the extraction, classification, and segmentation of phase-contrast MRI images. It enables rapid, accurate flow quantification for five blood vessels in a single ventricle registry without manual input.
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