How low can we go? The effect of acquisition duration on cardiac volume and function measurements in free-running cardiac and respiratory motion-resolved five-dimensional whole-heart cine magnetic resonance imaging at 1.5T

Background: Cardiovascular magnetic resonance (CMR) is the gold standard for assessing cardiac volumes and function using two-dimensional (2D) breath-held cine imaging. This technique, however, requires a reliable electrocardiogram (ECG) signal, repetitive breath-holds, and the time-consuming and proficiency-demanding planning of cardiac views. Recently, a free-running framework has been developed for cardiac and respiratory motion-resolved five-dimensional (5D) whole-heart imaging without the need for an ECG signal, repetitive breath-holds, and meticulous plan scanning. In this study, we investigate the impact of acquisition time on cardiac volumetric and functional measurements, when using free-running imaging, compared to reference standard 2D cine imaging. Methods: Sixteen healthy adult volunteers underwent CMR at 1.5T, including standard 2D breath-held cine imaging and free-running imaging using acquisition durations ranging from 1 to 6 min in randomized order. All datasets were anonymized and analyzed for left-ventricular end-systolic volume (ESV) and end-diastolic volume (EDV), as well as ejection fraction (EF). In a subset of data, intra- and inter-observer agreement was assessed. In addition, image quality and observer confidence were scored using a 4-point Likert scale. Finally, acquisition efficiency was reported for both imaging techniques, which was defined as the time required for data sampling divided by the total scan time. Results: No significant differences in left-ventricular EDV and ESV were found between free-running imaging for 1, 2, 3, 5, and 6 min and standard 2D breath-held cine imaging. Biases in EDV ranged from −2.4 to −7.4 mL, while biases in ESV ranged from −3.8 to 2.1 mL. No significant differences in EF were found between free-running imaging of any acquisition duration and standard 2D breath-held cine imaging. Biases in EF ranged from −2.8% to 0.94%. Both image quality and observer confidence in free-running imaging improved when the acquisition duration increased. However, they were always lower than standard 2D breath-held cine imaging. Acquisition efficiency improved from 13% for standard 2D cine imaging to 50% or higher for free-running imaging. Conclusion: Free-running CMR with an acquisition duration as short as 1 min can provide left-ventricular cardiac volumes and EF comparable to standard 2D breath-held cine imaging, albeit at the expense of both image quality and observer confidence.