We describe a model-based approach for real-time cardiac imaging
("fluoroscopy") that does not require cardiac gating. Unlike previous methods, the current approach produces a cine in which each reconstructed heartbeat is temporally resolved, rather than being a composite or weighted average of multiple actual heartbeats. The model used in the method captures the spatial and temporal-spectral characteristics of the heart. The model parameters are estimated as part of the MRI experiment and drive both data acquisition and cine reconstruction algorithm. The approach relies on a formulation of Dynamic MRI as a time-sequential sampling (TSS) problem, where only one sample in k-space can be acquired at any one time. The TSS theory provides a design for minimum redundancy data acquisition and reconstruction optimized for the dynamic object being imaged. The approach can produce a high temporal and spatial resolution movie of the heart, with multifold reduction in acquisition rate requirements compared to conventional methods. The method is demonstrated by simulation and in-vivo experiments. An extension of the method enables ungated free-breathing real-time cardiac imaging, by accounting in the model for both cardiac and respiratory motion during imaging.
Work with: Nitin Aggarwal, Qi Zhao, Saptarshi Bandyopadhyay
Audio (MP3 File, Podcast Ready)
Copyright. All Rights Reserved.