1339459 - Investigation of normative 3D fetal cardiovascular hemodynamics with 4D flow MRI

Echocardiography is critical to understand fetal cardiovascular physiology, but it is limited by fetal position and operator expertise. MRI can provide comprehensive assessment of blood flow independent of fetal position using carefully placed individual 2D phase contrast (PC) planes in each major vessel [1]. Recently, an iteration of the slice-to-volume approach, acquiring multi-planar, multi-slice 2D PC MRI reconstructed in 3D, quantified blood flow in 7 fetuses [2]. Here, we use Doppler ultrasound gated 4D flow MRI to visualize and quantify fetal cardiovascular hemodynamics.

Methods:

Ten healthy pregnant volunteers (gestational age=35±2 weeks) were prospectively recruited and scanned at 1.5T (n=2) or 3T (n=8) (Philips). Fetal cardiac gating was obtained with a MR-compatible Doppler ultrasound device (Northh) [3,4] positioned superficially over the fetal heart. Localizers covering the entire fetus were segmented to determine fetal mass [5]. 4D flow data encompassing the fetal heart and major vessels were acquired with resolution of 2.5 mm isotropic, SENSE=2×1.5 (phase × slice), and scan time ~3 min. Using EnSight (Ansys), flow was quantified in 3 adjacent planes in 6 vascular territories (Fig 1). Repeatability across planes was assessed via the within-subject coefficient of variation (WS-CV), and the mean was normalized by fetal mass to obtain flow in mL/min/kg, or by combined ventricular output (CVO=main pulmonary artery (MPA)+ascending aorta (AAo) flow) to compute relative segmental flow.

Results:

The mean±SD fetal mass was 2379±371 g and fetal heart rate was 136±9 bpm. Fig 2 shows 4D flow pathlines in the fetus of a volunteer and Fig 3 reports blood flow in the major vessels. CVO was 244±37 mL/min/kg, and in all segments, WS-CV< 20%. As expected, flow decreases from AAo to arch to isthmus due to branching head vessels, and pulmonary artery blood flow (PBF=MPA-ductus arteriosus (DA)) was low due to high pulmonary vascular resistance. In the descending aorta (DAo), flow increased post ductal and aortic arch convergence.

Conclusion:

We demonstrate the utility of 4D flow MRI for evaluation of the fetal circulation. Importantly, these data were acquired in ~3 min (vs 14-30 min for prior approaches [1,2]) and reconstructed online. In the presence of substantial fetal motion, 4D flow data can be corrupted, however with this method, it leads to loss of the gating signal and suspension of data collection as well.

In each major vessel, relative flow distributions agree within 10% of the 2D PC reference ranges reported in [1], though absolute flow is lower by 67-109 mL/min/kg. Comparison to [2] is less straightforward: flow is higher in some vessels while lower in others, but overall quantitative results are within 12-55 mL/min/kg of each other. Given the small vessel sizes, quantitative flow measurements are likely underestimated due to partial voluming, however the clinically important ability to evaluate fetal blood flow directionality and circulatory distributions is retained.