Meeting Banner
Abstract #4876

Highly Accelerated Myocardial Perfusion Using Physics-guided Deep Learning With Structure-encoded Coil Maps

Omer Burak Demirel1,2, Burhaneddin Yaman1,2, Steen Moeller2, Sebastian Weingärtner3, and Mehmet Akçakaya1,2
1Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, United States, 2Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, 3Department of Imaging Physics, Delft University of Technology, Delft, Netherlands

Synopsis

Myocardial perfusion cardiac MRI is widely used to functionally assess coronary artery disease. Although numerous acceleration techniques are used, improved spatio-temporal resolutions and coverage are desirable. Deep learning (DL) reconstruction has shown improvement over conventional reconstruction techniques at higher accelerations. Yet, SNR/contrast changes across perfusion dynamics hinder their generalization performance. In this work, we propose a multi-coil encoding operator that uses coil maps encoding structural information for physics-guided DL. This provides a uniform contrast at the network output across dynamics, leading to improved image quality compared to physics-guided DL with ESPIRiT coil maps, as well as conventional acceleration methods.

This abstract and the presentation materials are available to members only; a login is required.

Join Here