Free Breathing Single-Beat Myocardial Late Gadolinium Enhancement Imaging

Motivation: To overcome the limitations of LGE imaging, including respiratory motion artifacts and lengthy scan times, while enhancing myocardial scar imaging. Goal(s): To develop a deep learning-based free-breathing single-beat LGE. Approach: Free-breathing single-beat low-resolution 2D LGE images are acquired and followed by resolution enhancement generative adversarial inline neural network (REGAIN) to enhance the spatial resolution. Each slice was acquired in a single beat, followed by one beat for signal recovery. The entire left ventricular dataset was acquired in 20 heartbeats. Results: REGAIN improved image sharpness and quality of single-beat 2D LGE acquired with 4.7-fold acceleration with spatial resolution of 1.5 × 5 mm2. Impact: A rapid single-beat 2D LGE imaging can reduce CMR scan time, increase patient comfort, and reduce sensitivity to breathing motion.