Fast and Feasible: Two-Minute k-Space and Time–accelerated Aortic Four-dimensional Flow MRI

See also the article by Bollache et al in this issue.

Christopher J. François, MD, is an associate professor of radiology in the cardiovascular imaging and thoracic imaging section at the University of Wisconsin School of Medicine and Public Health. He is chief of the cardiovascular imaging section and director of the clinical MRI fellowship. His most current research includes 4D flow MRI in congenital heart disease, pulmonary hypertension, aortic dissection, portal hypertension, and pregnancy; exercise stress cardiac MRI; rapid quantitative cardiac MRI; and 3D printing of patient specific models. He is an associate editor for Radiology: Cardiothoracic Imaging.

Introduction

Four-dimensional (4D) flow-sensitive MRI is increasingly used clinically and in investigational studies to quantify blood flow and assess changes in hemodynamics in a variety of cardiovascular diseases. 4D flow MRI simplifies the acquisition of flow-sensitive MRI because a large, standard volume is acquired in all patients, with orientation of the analysis planes performed offline, after patients have left the scanner. This improves efficiencies and workflow, mitigating the need for supervising physicians to be present at the scanner throughout the examination. This is particularly true in congenital heart disease and valvular heart disease, when quantification of flow orthogonal to the direction of flow is critical and challenging. Although 4D flow MRI has been shown to reduce scan time for flow quantification in complex congenital heart disease, the relatively long acquisition times have impeded greater adoption in other clinical applications.

Several strategies for accelerating image acquisition have been applied to 4D flow MRI in the heart and aorta, including, but not limited to, the use of parallel imaging (1), echo-planar imaging (2), compressed sensing (3), and non-Cartesian acquisitions (4). Scan times on the order of 8–12 minutes have been achieved with these approaches. More recently, k-space and time (k-t) acceleration approaches have been used to reduce scan times even further (5). A potential disadvantage of using k-t acceleration methods is the temporal view sharing that could result in temporal blurring. The temporal view sharing may not have a significant effect on flow velocities or volumes. However, the impact on more advanced hemodynamic parameters, such as pulse wave velocity (PWV) and wall shear stress (WSS) that are highly dependent on temporal and spatial resolution, respectively, is incompletely understood.

In the study by Bollache et al, 4D flow MRI acquisition was reduced from approximately 12 minutes using generalized autocalibrating partially parallel acquisition (GRAPPA) (acceleration factor of 2) and respiratory navigator motion correction to 2 minutes using fivefold k-t acceleration without respiratory gating (6). This represents a dramatic improvement in efficiency that should translate into greater adoption of 4D flow MRI clinically. This is particularly true for flow velocity. The differences in peak systolic velocity between the GRAPPA accelerated acquisition and the k-t accelerated acquisition were insignificant and less than 10 cm/sec. This difference is unlikely to be clinically relevant.

The authors did observe significantly lower net flow through the ascending aorta than the left ventricular stroke volume in the subjects for whom standard cine cardiovascular MRI was available as a standard of reference. This is consistent with data from previous studies comparing 4D flow MRI flow to ventricular stroke volumes. Some of these differences can be explained by the prospective electrocardiographic triggering used for the 4D flow MRI acquisition, which results in missing data from the first part of systole.

The authors also reported results of WSS and PWV measurements. In general, the differences in WSS were not significantly different between the two acceleration methods. Significant, but small, differences in ascending aorta WSS were observed in patients. The differences in PWV between the two 4D flow MRI sequences were significant and slightly greater than 1 m/sec. Although differences in these more advanced hemodynamic parameters were observed, it is encouraging that the authors continued to observe differences in these parameters between healthy control subjects and patients in whom hemodynamic parameters would be expected to be abnormal. More work is still required to understand the impact these small differences in WSS and PWV would have on their use for research or clinical studies.

Currently, the greatest clinical use of MRI flow remains for quantifying flow velocity and volume. The introduction of a reproducible, accurate 4D flow MRI sequence that could be used to quantify these clinically relevant parameters in 2 minutes should lead to shorter scan times and simplified acquisitions, which combined should result in greater access to cardiovascular MRI. A benefit of using 4D flow MRI is the increased confidence in the flow measurements because it is straightforward to assess the consistency of the measurements by analyzing the flow at multiple locations in the volume. A major reason that 4D flow MRI has not been adopted more widely clinically is the relatively long scan times. With k-t acceleration methods resulting in much shorter acquisition times, the barrier to greater use of 4D flow MRI has been lowered.

There is still work to be done to confirm the consistency of the measurements using these highly accelerated acquisitions. For example, the effects of k-t acceleration techniques on time derivatives of flow are incompletely understood, including WSS and PWV. In addition, although WSS and PWV have been found in numerous research studies to be important discriminators between those without and those with disease, these have not yet been integrated into routine clinical care. Now that these parameters can be calculated from a 2-minute scan, will these indexes become part of a “routine” clinical cardiovascular MRI study?

In conclusion, Bollache et al should be commended for their work in accelerating 4D flow MRI acquisition times. As cardiovascular MRI becomes simpler and faster to perform, its use in the evaluation of patients with cardiovascular disease will increase. The ability to obtain 4D flow MR images in less than 2 minutes is an important step in that direction.