Abstract #0524

SAM-driven MaskNet for Left Ventricle Segmentation on Cine DENSE with Unsupervised Domain Adaptation

Siyue Li¹, Shu-Fu Shih¹, J. Paul Finn^1,2, Dan Ruan^2,3, Kim-Lien Nguyen^1,2,4,5, and Xiaodong Zhong^1,2

¹Department of Radiological Sciences, University of California Los Angeles, Los Angeles, CA, United States, ²Physics and Biology in Medicine Graduate Program, University of California, Los Angeles, CA, United States, ³Department of Radiation Oncology, University of California Los Angeles, Los Angeles, CA, United States, ⁴Division of Cardiology, VA Greater Los Angeles Healthcare System, Los Angeles, CA, United States, ⁵Division of Cardiology, University of California Los Angeles, Los Angeles, CA, United States

Synopsis

Keywords: Segmentation, Segmentation, Unsupervised Domain Adaptation, Left Ventricle Segmentation, DENSE

Motivation: This study leverages unsupervised domain adaptation (UDA) and the Segment-Anything-Model (SAM) for automated and accurate left ventricle (LV) segmentation of displacement encoding with stimulated echoes (DENSE) MRI, enabling reproducible myocardial strain quantification.

Goal(s): Develop a UDA-based framework driven by SAM, to investigate its LV segmentation accuracy and reproducibility in strain analysis using DENSE MRI.

Approach: MaskNet leverages UDA to transfer segmentation knowledge from cine steady-state free precession (SSFP) images to DENSE images and SAM to generate robust masks. Strain results using MaskNet-derived contours were validated against manual contours.

Results: MaskNet demonstrated accurate LV segmentation and reproducible strain results.

Impact: This research advances automated LV segmentation in DENSE MRI by integrating UDA and foundation models, leveraging cine SSFP images and established annotations to improve deep learning model performance despite limited availability of specialized MR images like DENSE.

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.

Click here for more information on becoming a member.