Abstract #4280

Physics-informed Model Selection for Robust Deep Learning Segmentation of Multi-center Perfusion CMR: Initial Findings from the SCMR Registry

Dilek M. Yalcinkaya^1,2, Arian M. Sohi², Khalid Youssef³, Luis Zamudio³, Michael Elliott⁴, Venkateshwar Polsani⁵, Rohan Dharmakumar³, Robert Judd⁶, Matthew Tong⁷, Dipan Shah⁸, Orlando Simonetti⁷, and Behzad Sharif^2,9

¹Electrical and Computer Engineering, Purdue University, West Lafayette, IN, United States, ²Laboratory for Translational Imaging for Microcirculation, Weldon School of Biomedical Engineering, Purdue University, Indianapolis, IN, United States, ³Krannert Cardiovascular Research Center, Indiana University School of Medicine, Indianapolis, IN, United States, ⁴Atrium Health Sanger Heart & Vascular Institute Kenilworth, Charlotte, NC, United States, ⁵Piedmont Heart Institute, Atlanta, GA, United States, ⁶Intelerad Medical Systems, Durham, NC, United States, ⁷The Ohio State University, Columbus, OH, United States, ⁸Houston Methodist DeBakey Heart and Vascular Center, Houston, TX, United States, ⁹Department of Radiology & Imaging Sciences, Indiana University School of Medicine, Indianapolis, IN, United States

Synopsis

Keywords: Analysis/Processing, Analysis/Processing, Cardiovascular, Myocardium, Perfusion

Motivation: Accurate segmentation of first-pass perfusion CMR is critical for reliable myocardial blood flow analysis.

Goal(s): To enable generalizaton to multi-center datasets.

Approach: Incorporating cardiac magnetic resonance physics-informed features in the deep learning model selection process.

Results: Our results, which leverage multi-center perfusion CMR datasets from the SCMR registry, suggest that combining physics-informed guidance with uncertainty-based deep learning model selection improves the segmentation performance in deep learning-based analysis of multi-center stress FPP datasets.

Impact: The proposed hybrid approach has the potential to improve the reliability of fully automated stress/rest FPP analysis in clinical settings and in multi-center clinical trials.

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.