Introduction

Magnetic resonance imaging of the cardiovascular system is considered a valuable diagnostic tool for studying congenital or development abnormalities in children and adults. However, simple, high-quality imaging of the fetal heart is challenging due to lack of direct in-utero cardiac gating. We aimed to employ a recently introduced Doppler ultrasound (DUS) device ^1-2 and optimized pulse sequences for structural and functional imaging of the fetal heart and to establish a standard acquisition approach for high-quality fetal cardiovascular MRI in the clinical practice.

Methods

All pregnant women with or without previously diagnosed or suspected heart malformations in prenatal ultrasound underwent fetal cardiovascular MRI on 3.0T or 1.5T clinical whole-body systems (Ingenia, Philips Healthcare, Best, the Netherlands) with 70 cm bore and a standard 32-channel torso coil. Subjects were examined in either supine or lateral decubitus position. A MR-compatible DUS device was used to sense the fetal cardiac motion and generate signals for MR image acquisition with direct fetal cardiac gating ^1-2. Imaging pulse sequences for cardiovascular structural and functional exams were adapted from conventional protocols and these included:

• Survey (i.e. localization or scout scan) based on multi-planar gradient-echo sequences without cardiac gating
• Morphological exam based on single-shot turbo-field echo with balanced steady-state free precession (bTFE or bSSFP) or single-shot turbo-spin echo (TSE) sequences, in orthogonal orientations
• Functional exam based on multi-shot bTFE cine sequences with cardiac gating
• 4D flow based on phase-contrast velocity-encoded turbo-field echo (TFE) sequences with cardiac gating
• Native myocardial T1 mapping based on Modified Look-Locker Imaging (MOLLI) with cardiac-triggered single-shot bTFE

Optimization was done for all cardiac-gated or triggered pulse sequences, in particular, for a trade-off between the temporal and spatial resolutions, as the fetal cardiac frequency was substantially higher than that in adults. In addition, scan times were kept reasonably short either to accommodate minimally repeated breathholds or to reduce the risk of abrupt fetal movement. Cine exams were investigated with 1) high spatial resolution 1.2 x 1.0 x 5 mm³ (temporal resolution 23-29 ms) and 2) high temporal resolution 14 to 16 ms (1.6 x 1.4 x 5 mm³) for maternal breathhold (BH) and free breathing (FB), respectively. 4D flow was acquired with a maximum velocity encoding (VENC) value of 120 cm/s. The recently introduced compressed SENSE (C-SENSE) technique ^3,4, which employed compressed sensing together with coil sensitivity information, was applied whenever available for scan acceleration. Main imaging parameters of the cardiac-gated pulse sequences were summarized in Table 1 and were kept comparable between 3.0T and 1.5T. Images were assessed visually according to general quality, artifacts, depiction of key anatomical structures and myocardial contraction. 4D flow data were post-processed using GTFlow (GyroTools, Zurich, Switzerland) for visualization and quantification.

Results and Discussion

Nine fetuses were studied (gestational age 30 to 36 weeks). Fetal cardiac frequency ranged from 130 to 180 beats per minute (bpm). All data were successfully acquired with DUS gating. Two patients were excluded from analysis due to severe fetal movement.
Figure 1 demonstrated a typical optimization process from single-shot non-gated bTFE without gating to multi-shot bTFE cine with DUS fetal cardiac gating for better delineation of the cardiac motions. Imaging times of functional cine exams were typically 7 to 12 s for BH scans and 30 to 40 s for FB scans with multiple averaging (NSA = 3 to 4) to mitigate gross motion without respiratory navigator. Different anatomical orientations including 4-chamber and short-axis cardiac views showed good image quality with excellent myocardium-to-blood contrast and clear fetal myocardial contractions in both sequence setups. Typical case examples on 1.5T and 3.0T were shown in Figure 1 and 2. No significant difference was found between BH and FB scans under shallow and regular maternal breathing. Otherwise, BH scans were deemed better with clearly less motion-induced aliasing artifacts.
Fetal circulation could be visualized in 4D flow, showing ductus arteriosus (DA) connecting the trunk of the pulmonary artery to the proximal descending aorta (AoD), which allows blood from the right ventricle to bypass the fetus's fluid-filled non-functioning lungs (Figure 3). Blood flow in the ductus was seen with high velocities, which mixed with the blood from the ascending aorta and created spiral-shape vortices that were also clearly visible in AoD. Quantitative flow measurements in AoD were in accordance to the previous reports ^3-4 (results not shown here). T1 maps could be obtained and ROI analysis from the inter-ventricular septum revealed a mean native myocardium T1 of 1090 ± 69 ms (Figure 4). In certain cases image registration may help to further compensate residual motion in post-processing, as in the adult case ⁵.

Conclusion

Fetal cardiovascular MRI was possible based on direct fetal cardiac gating from the DUS device. The benefits have been demonstrated with promise in high-quality morphological and functional studies of the fetal heart using routinely available imaging techniques. Preliminary experience in patients with free breathing may further foresee a wide clinical adoption for evaluation of congenital pathologies. Future studies are warranted to investigate its robustness and clinical performance.

Acknowledgements

The authors thank Dr. Hendrik Kooijman-Kurfuerst for his help in preparation of the scan protocols.