4 - Measuring Ejection Fraction in Reformatted 5D Image Data Acquired with Free-Running Whole-Heart MRI

Background: Ejection fraction (EF) is an essential component of the clinical cardiac magnetic resonance (MR) exam. Typically, EF is found through multiple 2D, ECG-gated, breath-hold cine acquisitions. Recent advancements have made it possible to acquire whole-heart volumes across multiple respiratory and cardiac phases, thereby allowing reconstruction of 5D image data (3 spatial dimensions, plus cardiac and respiratory phases) [1]. This methodology has been developed into a fully self-gated, free-running framework (FRF) [2] that has been optimized primarily for anatomic great vessel and coronary artery imaging. We hypothesized that multiplanar reformation (MPR) of 5D images obtained with FRF and reconstructed at end-expiration can provide images adequate for calculating EF values which agree with those calculated through reference-standard 2D multi-slice breath-hold method.

Methods: Images were acquired in 6 patients (mean age of 42 ± 14; 3 female) who had clinically ordered cardiac MR examinations at Emory University Hospital on a 1.5T Siemens MAGNETOM Avanto fit Scanner. For each patient, the clinical standard stack of 2D, ECG-gated SSFP short-axis cines were acquired across the LV during breath-hold. EF was calculated by manual tracing of LV endocardial borders using the program Segment. Additionally, data were collected with FRF that incorporates 1) a free-running acquisition with a 3D spiral phyllotaxis trajectory and SSFP readout [3,4] as well as 2) a fully automated cardiac and respiratory signal extraction for data binning and 3) a compressed sensing reconstruction algorithm [2].  Approximately 110,000 radial lines were retrospectively binned into ~25 cardiac and 4 respiratory phases, leading to ~100 motion-resolved 3D whole-heart image volumes across the cardiac and respiratory cycles. A quantitative comparison of EF was done between the 2D breath-hold ECG-gated sequence and the reformatted 5D dataset.

Results: The reference-standard 2D technique gave a mean EF of 51% ± 16% while the 5D FRF technique gave 51% ± 13%. The two groups showed good correlation and agreement [Figure 2].

Conclusion: This study showed that 5D images obtained with the fully self-gated free running framework can be reconstructed and reformatted to determine EF, and values are statistically equivalent to values calculated using the clinical reference standard technique of a series of 2D, ECG-gated, breath-hold images. , indicating that the technique can be implemented in a typical patient group. Implementation of this technique would allow clinicians to use a single, free-breathing, non-ECG gated sequence with minimal planning to provide the data necessary for evaluation LV volumes and EF, thereby allowing for a more time-efficient clinical workflow.