Meeting Banner
Abstract #4278

Co-Evolution of CNNs and Learning Environments for Automated Post-Processing of T2 Parametric Mapping in Cardiovascular Magnetic Resonance

Thomas Hadler1,2, Clemens Ammann2,3, Philine Reisdorf1,2, Steffen Lange4, and Jeanette Schulz-Menger1,2
1Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany, Berlin, Germany, 2Working Group on CMR, Experimental and Clinical Research Center, a cooperation between the Max Delbrück Center for Molecular Medicine in the Helmholtz Association and Charité – Universitätsmedizin Berlin, Berlin, Germany, Berlin, Germany, 3Department of Cardiology and Nephrology, HELIOS Hospital Berlin-Buch, Berlin, Germany, Berlin, Germany, 4Department of Computer Sciences, Hochschule Darmstadt - University of Applied Sciences, Darmstadt, Germany, Darmstadt, Germany

Synopsis

Keywords: Analysis/Processing, AI/ML Software, Cardiovascular, Quantitative Imaging, Myocardium

Motivation: Training robust convolutional neural networks (CNNs) for cardiac magnetic resonance (CMR) imaging is essential, as varying imaging protocols and patient demographics can significantly affect performance.

Goal(s): The aim is to develop an evolutionary algorithm that computes learning environments (LEs) for CNNs, enhancing their adaptability for myocardial segmentation and reference point estimation in T2 parametric mapping.

Approach: The algorithm trains 48 CNNs simultaneously, adjusting LEs through mutation based on performance evaluations.

Results: Performance improved over time, achieving a Dice score of 86% and a reference point distance of 6mm, indicating effective adaptability across varying cardiac geometries and pixel intensities.

Impact: The evolutionary algorithm improves CNN performance in cardiovascular MRI by dynamically optimizing learning environments, enhancing adaptability across varied imaging conditions. This approach strengthens model generalizability and reliability, making it a promising method for advancing robust AI tools in clinical practice.

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.

Keywords