Abstract:Four-dimensional (4D) flow assessment by MRI is becoming an attractive tool to assess hemodynamics in different applications of cardiac chambers, valves, and large thoracic and abdominal vessels as well as in neurovascular applications. Cardiac valvular flow evaluation by 4D flow has unique characteristics that differ from echocardiography and enable image acquisition without geometric assumptions, flow measurements with plane reformatting that account for valve plane motion during the cardiac cycle, simultaneous analysis of flow across all four valves, and advanced fluid biomechanics measurements such as turbulent kinetic energy, viscous energy loss rate, and wall shear stress. It is also advantageous for its simplicity in direct regurgitant volume assessment when compared with the conventional MRI method of indirect measurement, which involves two volumetric measurements and their subtraction. Good to excellent correlations between the 4D flow and conventional MRI techniques are reported in the literature. Despite these advantages, 4D flow image is a relatively novel technique and its image acquisition is still timeconsuming (up to 15 minutes) even when utilizing acceleration techniques that represent hurdles for clinical adoption. The SCMR consensus statement recommends parallel imaging (PI) as an image acquisition acceleration technique, but other approaches including compressed sensing (CS), nonCartesian trajectory techniques, and artificial intelligencebased cardiac MRI reconstruction have also been introduced. Examples of such approaches include k-adaptive-t autocalibrating reconstruction for Cartesian sampling (kat-ARC) and pseudospiral Cartesian sampling with random undersampling in time (PROUD) combined with CS reconstruction. The combination of acceleration techniques may effectively reduce scanning time but can introduce other impacts in terms of susceptibility to artifacts, image quality, and consistency of intra-cardiac flow quantifications. In this issue of Journal of Magnetic Resonance Imaging, Emrich et al studied the highly accelerated CS-based image acquisition in 4D flow imaging and validated it in comparison to the conventional technique of PI-based GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) for intra-cardiac valve flow assessment in 50 healthy volunteers in a prospective study. CS 4D flow data were undersampled using a spiral phyllotaxis subsampling pattern and the acceleration factors were set in 3 for GRAPPA and 7.7 for CS. All four intra-cardiac valve flow measurements and the diastolic function assessment of E/A and E/e’ were included in the validation effort. A significant acquisition time efficiency of 44% was achieved in CS-based acquisition but it underestimated flow with an average error of 17%. The underestimation was less than 10% for large forward volumes and net forward volumes. The authors conclude that highly accelerated 4D flow acquisition is feasible despite volume underestimation. The current study sheds light on the fact that optimal image acceleration is selected on the basis of a balance between scan time clinical feasibility and ideal data quality. The CS-based acceleration method is efficient in 4D flow imaging, considering its high-dimensional characteristics. Meanwhile, the possible underestimation associated with the highly accelerated image acquisition is a concern as presented in the current study. When utilizing a CS-based high acceleration rate of 7.7, the aortic 4D flow image scan time can be shortened substantially up to 2 min with the trade-off being the underestimation of peak velocities and peak flow generally within 13% when compared with the conventional PI-based methods. In the pediatric population, for example, the short scan time is a great advantage but underestimation needs to be avoided when a much smaller flow volume needs to be measured compared with adults. Accordingly, selecting the appropriate image acceleration strategy that fits the clinical situation is essential. The underestimation of the 4D flow is influenced by the temporal undersampling. In this context, temporal sparsity modification with increased incoherence may improve data quality, although its feasibility in 4D flow imaging needs further investigation. Some other remaining technical challenges are recognized such as temporal and spatial resolutions for the fine anatomy of coronary arteries and left atrial appendages, for

Inline automatic quality control of 2D phase‐contrast flow MRI for subject‐specific scan time adaptation

Reduced Field of View Single‐shot Spiral Perfusion Imaging

Rotated spiral RARE for high spatial and temporal resolution volumetric arterial spin labeling acquisition

Rapid 3D Dynamic Arterial Spin Labeling with a Sparse Model-Based Image Reconstruction

Highly accelerated free-breathing real-time 2D flow imaging using compressed sensing and shared velocity encoding

Quality Control for 4D Flow MR Imaging

Automated Quantification of Simple and Complex Aortic Flow Using 2D Phase Contrast MRI

Combined free-running four-dimensional anatomical and flow magnetic resonance imaging with native contrast using Synchronization of Neighboring Acquisitions by Physiological Signals

Real-time Phase-Contrast Flow Cardiovascular Magnetic Resonance with Low-Rank Modeling and Parallel Imaging

bSSFP Phase Contrast (PC-SSFP) at 0.55T Applied to Aortic Flow

Coronary Flow Assessment Using Accelerated 4D Flow MRI With Respiratory Motion Correction

Fully Self-Gated Whole-Heart 4D Flow Imaging from a Five-Minute Scan

Background phase induced steady‐state effects in velocity quantification using phase‐contrast MRI

4D Real-Time Phase-Contrast Flow MRI with Sparse Sampling

Learning-Based Quality Control for Cardiac MR Images

Comparison of Echo-Planar Imaging and Compressed Sensing in the Estimation of Flow Metrics from Aortic 4D Flow MR Imaging: A Healthy Volunteer Study

Fully Automated Valve Segmentation for Blood Flow Assessment From 4D Flow MRI Including Automated Cardiac Valve Tracking and Transvalvular Velocity Mapping

Editorial for “Highly Accelerated Compressed Sensing 4D Flow for Intracardiac Flow Assessment”

A deep learning algorithm for noise reduction in a novel respiratory-triggered single-shot phase sensitive inversion recovery myocardial delayed enhancement cardiac MRI pulse sequence

Automatic measurement plane placement for 4D Flow MRI of the great vessels using deep learning

Highly efficient free-breathing 3D whole-heart imaging in 3-min: single center study in adults with congenital heart disease