4060
Multislice and Multidimensional Blood Flow Mapping in Fetal Congenital Heart Disease Using Highly Undersampled Phase-Contrast MRI1Medical Biophysics, University of Toronto, Toronto, ON, Canada, 2Translational Medicine, Hospital for Sick Children, Toronto, ON, Canada, 3Cardiology, Hospital for Sick Children, Toronto, ON, Canada, 4Radiology, Lausanne University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland, 5Paediatrics, University of Toronto, Toronto, ON, Canada
Synopsis
In this study, multislice and multidimensional fetal blood flow mapping was performed using highly undersampled radial phase contrast MRI. The goal of this work was to visualize intracardiac flow in human fetuses with congenital heart disease, where complex flow patterns are difficult to capture with conventional, 2D phase contrast techniques. Reconstructions were performed using compressed sensing, with retrospective motion correction and image-based cardiac gating. The developed technique allowed whole fetal heart coverage in reasonable time and provides insight into multi-directional intracardiac flows. We present results from 5 human fetuses with congenital heart disease and demonstrate the first multi-directional intracardiac flows obtained by MRI.
Introduction
PCMR has previously enabled quantification of through-plane fetal blood flow in pregnancies with fetal congenital heart disease (CHD) [1]. Unfortunately, fetal motion during the scan leads to spatial offsets between localisers and phase contrast acquisitions, such that slice planning must be repeated. The overall scan time is thus increased at the expense of maternal discomfort, while non-diagnostic information is acquired during repeated localisers. Recent advances in accelerated multislice cardiac MRI with motion compensation has allowed visualization of the fetal heart in cases of CHD [3,4], and extension of these methods to multidimensional phase contrast (PC) MR has been demonstrated in healthy fetuses [5]. In this current work, we use highly accelerated radial PCMR for multislice, multidimensional flow imaging in fetuses with CHD to examine intracardiac flows. This approach involves rejection of gross fetal motion, correction for in-plane translational motion, cardiac gating using metric optimized gating, and compressed sensing (CS) to reconstruct multidimensional fetal flow CINE images. The feasibility of this approach is tested in five human pregnancies with CHD. We present here the first multislice and multidimensional PCMR flow and anatomical CINE images of these fetal hearts.Methods
Acquisition: All scans were performed as part of an ethically approved study, and informed, written consent was obtained from all participants who were referred to MR after CHD was diagnosed through echocardiography. A previously described PCMR sequence with golden-angle radial trajectory and multidimensional velocity encoding [5] was used to acquire data in third trimester pregnancies (32-36 weeks) with fetal CHD. A summary of study participants (gestational age, CHD diagnosis) is provided in Table 1.
Imaging was performed using a 1.5T MRI system (MAGNETOM AvantoFIT, Siemens Healthcare, Erlangen, Germany) and acquisition parameters included: total spokes=1600 (across all flow encodes), field-of-view=256x256mm2, resolution=1x1x4mm3, TR=5.5ms, scan time=8.8s/slice, VENC=80cm/s. 10-13 slices were used to cover the fetal heart, yielding scan times of 1.4-1.9min. Initial reconstructions were performed online, generating time-averaged anatomical and multidimensional flow images to confirm slice prescriptions and plan subsequent scans.
Reconstruction (offline): First, real-time images were reconstructed for motion compensation using a 64-spoke sliding window (Figure 1A). Reconstruction was performed using CS with spatial total variation (STV) and temporal total variation (TTV) constraints, with λ=0.05 and 0.025, respectively. Through plane motion in each slice was rejected automatically from the pipeline using outlier rejection (mutual information), while in-plane translational motion was corrected by rigid image registration.
Next, for cardiac gating, the motion-corrected data were reconstructed to create another real-time series but at higher temporal resolution (8-spoke sliding window), using the same CS algorithm (Figure 1B). Metric optimised gating was performed by minimizing the spatiotemporal entropy over a region encompassing the fetal heart.
Finally, CINE reconstructions using the retrospectively gated and motion-corrected data were performed using CS with STV, TTV and complex difference constraints, with λ=0.05, 0.05 and 0.025 respectively and view sharing over one half cardiac bin (Figure 1C, 1D).
Velocity and magnitude images across all subjects and slices were evaluated by visual inspection. Additionally, in one subject, a 4D flow vector visualization was generated (Siemens 4D Flow) [6], following spatial (mutual information) and temporal (entropy) registration of the acquired slices.
Results
Anatomical (magnitude) and velocity CINE reconstructions of one slice from each subject are shown in Figure 2. Qualitatively, they show altered cardiac structures and anomalous flows consistent with diagnosis as in Table 1. P1 demonstrates thickened myocardium (biventricular dilation) and vortex flow formation in the left ventricle. In P2 and P3 tricuspid regurgitation appears as signal loss in anatomical images and rapid flow reversals in velocity images. P4 shows moderate hypoplastic left ventricle with reduced flow into the right ventricle. P5 has severe right ventricular hypoplasia with reduced flow into the left ventricle. Figure 3 shows five contiguous slices from subject P1, descending from the base of the heart. Evidence of aortic regurgitation is observed in the top 3 slices and flow vortices are visible in the lower 2 slices. These observations are more evident in the planar vector plots provided in the 4D flow visualization shown in Figure 4.Discussion
In this work we present the first multislice and multi-dimensional velocity images of the fetal heart. Our method achieves complete cardiac coverage in less than 2 minutes. By providing both anatomical and flow data, the acquisition yields comprehensive, crucial diagnostic information. The resulting visualizations of the highly complex intracardiac flows will support enhanced understanding of CHD and provide valuable information for surgical planning and evaluation of treatment.Conclusion
We have demonstrated that highly accelerated radial PCMR can be used to image intracardiac flows with whole heart coverage in the fetuses with CHD. The initial feasibility of the strategy was tested in five pregnancies with CHD.Acknowledgements
No acknowledgement found.References
1. Sun, L., Macgowan, C. K., Sled, J. G., Yoo, S. J., Manlhiot, C., Porayette, P., ... & Hickey, E. (2015). Reduced fetal cerebral oxygen consumption is associated with smaller brain size in fetuses with congenital heart disease. Circulation, 131(15), 1313-1323.
2. Roy CW, Seed M, Kingdom JC, Macgowan CK. Motion compensated cine CMR of the fetal heart using radial undersampling and compressed sensing. J Cardiovasc Magn Reson. 2017;19:29.
3. van Amerom, J. F., Lloyd, D. F., Price, A. N., Kuklisova Murgasova, M., Aljabar, P., Malik, S. J., ... & Hajnal, J. V. (2018). Fetal cardiac cine imaging using highly accelerated dynamic MRI with retrospective motion correction and outlier rejection. Magnetic resonance in medicine, 79(1), 327-338.
4. Goolaub, D.S., Roy, C., Schrauben, E., Sussman, D., Marini, D., Seed, M., Macgowan, C. Multidimensional Fetal Flow Imaging with MRI: A Feasibility Study. JCMR. (In press)
5. Gulsun MA, Jolly MP, Guehring J, Guetter C, Littmann A, Greiser A, et al. A novel 4D flow tool for comprehensive blood flow analysis International Society for Magnetic Resonance in Medicine Melbourne, Victoria, Australia 2012.
Figures
