Fast high Spatiotemporal-Resolution Spiral bSSFP Cardiac Cine with Disentanglement-based Reconstruction and Segmentation

Motivation: MR-Imaging of rapid processes, such as the beating heart, typically relies on segmented acquisitions. Improving temporal resolution often requires extended acquisition windows, potentially increasing the burden on patients. Goal (s): A highly-efficient spiral bSSFP acquisition scheme, combined with a deep learning-based reconstruction is employed to reduce acquisition times for high-temporal-resolution cine imaging. Approach: Spiral acquisitions were performed on a subject with regular heartbeat, enabling creation of segmented, undersampled cine frames with temporal resolutions of 3. 7\, ms, 7. 4\, ms and 11. 1\, ms. Additionally, real-time frames were acquired with a temporal resolution of 48. 1\, ms. Results: The proposed approach achieves improved accuracy in depicting cardiac motion compared to real-time imaging. Impact: With undersampled spiral bSSFP acquisitions and a disentanglement-based reconstruction method high temporal resolution cardiac cine imaging can be acquired with up to 3. 7\, ms/frame, potentially improving accuracy in left ventricular strain analysis and cardiac function assessment in subjects with regular heartbeat.