4072

T1 and T2 mapping of the placenta and fetal brain with magnetic resonance fingerprinting (MRF) during maternal hyperoxia

Jeffrey N Stout¹, Congyu Liao², Esra Abaci Turk¹, Borjan Gagoski¹, P. Ellen Grant¹, Lawrence L. Wald², and Elfar Adalsteinsson^3,4

¹Fetal-Neonatal Neuroimaging Developmental Science Center, Boston Children's Hospital, Boston, MA, United States, ²Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, United States, ³Harvard-MIT Health Sciences and Technology, IMES, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁴Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States

Synopsis

Quantitative MRI may improve diagnosis and monitoring of placental disease by characterizing baseline oxygen content and dynamic oxygen transport during hyperoxia. MRF permits fast quantitative imaging of the adult brain and heart, but use during pregnancy targeting the placenta and fetal brain presents challenges due to motion and large B₁⁺ variation. We tested the accuracy of MRF in phantoms and then scanned pregnant mothers, estimating T₁ and T₂ of the placenta and fetal brain. MRF-based T₁ and T₂ mapping is a promising technique to determine placental oxygenation at baseline and oxygenation changes in the placenta and fetal brain after hyperoxia.

Introduction

Noninvasive methods to diagnose and monitor placental function are needed. MRI has the potential to resolve physiology related to oxygen transport through the placenta to the fetal brain, but faster quantitative mapping techniques are needed to freeze maternal and fetal motion, and to capture time-varying physiology. Placental Δ(1/T₂*) or ΔR₂* mapping during maternal normoxia-hyperoxia protocols is sensitive to underlying placental disease,¹ but diagnoses using ΔR₂* alone are hampered by large signal variations between subjects.² T₁ and T₂ are sensitive to placental physiology³ and blood oxygenation,⁴ but fast and robust quantitative techniques are needed. We implemented slice profile and B₁⁺ corrected MRF capable of producing T₁ and T₂ maps in nine seconds per slice, and tested it on phantoms to determine accuracy in the pregnant abdomen. We then acquired relaxivity maps during maternal breath hold, at baseline and during hyperoxia, to determine T₁ and T₂ of the placenta and fetal brain. Fast T₁ and T₂ mapping methods are a critical next step toward understanding baseline placental physiology and oxygen transport.

Methods

Ten pregnant women (mean gestational age (GA) 32.6 weeks) were scanned with IRB approval on a Siemens 3T scanner while breathing room air and after breathing 100% oxygen for 5 or 10 minutes. Validation scans were performed on the National Institute of Standards and Technology (NIST) MRI System Phantom and an anthropomorphic 22-week fetal phantom,^5,6 with spin echo imaging with and without inversion recovery to determine ground truth T₁ and T₂, respectfully. Single-slice inversion-recovery FISP MRF⁷ was used with acquisition parameters shown in Figure 1, resolution=3x3x4mm³, field of view=390mm, echo time=2.7ms, acquisitions=720, total time (TA)=9s during maternal breath hold. Pre-saturation turbo-flash (TA=30s) was used to estimate B₁⁺ maps, which were smoothed to compensate for motion corruption by fitting with a 6^th-order polynomial (Figure 2). T₁ and T₂ parameter maps were determined by the sliding window method.⁸ B₁⁺ and slice profile correction were performed by approximating the slice profiles for all excitations as a scaled versions of the profile for the maximum achieved flip angle for each B₁⁺ value, and then using B₁⁺ as prior information during parameter matching.^9,10 Regions of interest (ROI) in the placenta and fetal brain were drawn manually.

Results

Figure 3 shows MRF phantom performance, and the corrections for B₁⁺ and slice profile non-idealities in the large B₁⁺ range we observed (e.g. 73% change across abdomen). Observed mean absolute errors for the relaxivity range of fetal tissue were 57 ms (4%) for T₁ and 18 ms (15%) for T₂. Example in vivo relaxivity maps are shown in Figure 4. Baseline placenta T₁=1770±173ms and T₂= 55±7ms (mean±st.dev.). Baseline brain T₁=2112±521ms and T₂=158±114ms, for N=6 subjects (mean GA=33.2weeks). Placental Δ(1/T₁) or ΔR₁ between normoxia and hyperoxia when subjects with contractions were excluded (N = 4) was -0.012±0.012s^-1. Figure 5 shows correlations between T₁, T₂ and ΔR₁ with GA (brain T₂ was statistically significant).

Discussion

Quantitative imaging to probe oxygen transport dynamics in pregnancy holds great promise, and we demonstrate one approach to fast, simultaneous T₁ and T₂ mapping using MRF. Our MRF implementation showed accuracy similar to previous studies.¹¹ The baseline trend in placental T₁ with GA and the overall mean T₁ agrees with the literature.^3,12 The significant correlation between fetal brain T₂ and GA is similar to previous reports of T₂* and GA.¹³ The observed mean negative placental ΔR₁, against the backdrop of several reports of positive ΔR₁,^14,15 requires further investigation. It could be due to physiological noise affecting our comparatively short (9 sec versus ~3 min) measurement, differences in sensitivity to the various blood compartments in the placenta since maternal arterial blood is the only compartment that theoretically should show a positive ΔR₁ with hyperoxia, or MRF measurement error. Next, we will make multiple MRF measurements at baseline and during hyperoxia to determine the magnitude of the physiological variations. With continued refinement of the MRF technique to improve its accuracy for fetal imaging via magnetization preparation,¹⁶ implementation of optimal TR/FA progression,¹⁷ and motion correction,¹⁸ parameter maps could permit a more detailed analysis of the partial pressure of oxygen, blood oxygen saturation and hematocrit in the placenta.

Conclusion

MRF holds promise for the quantitative imaging of the placenta and fetal brain, and may lead to metrics of placental physiology and oxygen transport dynamics from the mother to the fetus. Work is needed to refine MRF for use in the pregnant abdomen by correcting for motion and by targeting relaxivity ranges not found in the adult brain.

Acknowledgements

NIBIB R01EB017337 “Advanced fetal imaging”, NICHD U01HD087211 “Noninvasive realtime assessment of placental structure and function with novel MR imaging methods”, and NIBIB R01EB017219, “Magnetic Resonance Fingerprinting (MRF) for Improved High Field MR.”

References

1. Luo, J. et al. In Vivo Quantification of Placental Insufficiency by BOLD MRI: A Human Study. Sci. Rep. 7, 3713 (2017).

2. Sinding, M. et al. Placental baseline conditions modulate the hyperoxic BOLD-MRI response. Placenta 61, 17–23 (2018).

3. Wright, C. et al. Magnetic resonance imaging relaxation time measurements of the placenta at 1.5 T. Placenta 32, 1010–1015 (2011).

4. Portnoy, S., Milligan, N., Seed, M., Sled, J. G. & Macgowan, C. K. Human umbilical cord blood relaxation times and susceptibility at 3 T. Magn. Reson. Med. 00, 1–13 (2017).

5. Keenan, K. et al. Multi-site, multi-vendor comparison of T1 measurement using ISMRM/NIST system phantom. in Proc. Int. Soc. Magn. Reson. Med. 3290 (2016).

6. Spatz, M., Garcia-Polo, P., Keil, B., Ha, C. & Wald, L. L. A 64 channel 3T array coil for highly accelerated fetal imaging at 22 weeks of pregnancy. in Proc. Intl. Soc. Mag. Reson. 25, 1220 (2017).

7. Jiang, Y., Ma, D., Seiberlich, N., Gulani, V. & Griswold, M. a. MR fingerprinting using fast imaging with steady state precession (FISP) with spiral readout. Magn. Reson. Med. 74, 1621–31 (2015).

8. Cao, X. et al. Robust sliding-window reconstruction for Accelerating the acquisition of MR fingerprinting. Magn. Reson. Med. 00, 1–10 (2016).

9. Buonincontri, G. & Sawiak, S. J. MR fingerprinting with simultaneous B1 estimation. Magn. Reson. Med. 76, 1127–1135 (2016).

10. Ma, D. et al. Slice profile and B1corrections in 2D magnetic resonance fingerprinting. Magn. Reson. Med. 78, 1781–1789 (2017).

11. Jiang, Y. et al. Repeatability of magnetic resonance fingerprinting T1 and T2 estimates assessed using the ISMRM/NIST MRI system phantom. Magn. Reson. Med. 1457, 1452–1457 (2016).

12. Ingram, E., Morris, D., Naish, J., Myers, J. & Johnstone, E. MR Imaging Measurements of Altered Placental Oxygenation in Pregnancies Complicated by Fetal Growth Restriction. Radiology 000, 162385 (2017).

13. Blazejewska, A. I. et al. 3D in utero quantification of T2* relaxation times in human fetal brain tissues for age optimized structural and functional MRI. Magn. Reson. Med. 00, (2016).

14. Ingram, E. et al. R1 changes in the human placenta at 3 T in response to a maternal oxygen challenge protocol. Placenta 39, 151–153 (2016).

15. Huen, I. et al. R1 and R2* changes in the human placenta in response to maternal oxygen challenge. Magn. Reson. Med. 70, 1427–1433 (2013).

16. Hamilton, J. I. et al. MR fingerprinting for rapid quantification of myocardial T1, T2, and proton spin density. Magn. Reson. Med. 77, C1 (2017).

17. Zhao, B. et al. Optimal Experiment Design for Magnetic Resonance Fingerprinting: Cramer-Rao Bound Meets Spin Dynamics. 1–18 (2017). doi:10.1109/TMI.2018.2873704

18. Mehta, B. B. et al. Image reconstruction algorithm for motion insensitive MR Fingerprinting (MRF): MORF. Magn. Reson. Med. 1–16 (2018). doi:10.1002/mrm.27227

Figures

Figure 1: The sinusoid variation in flip angles (FA) between 3-78° and the perlin noise pattern variation in repetition times (TR) between 12-12.81 ms used in the FISP based MRF sequence (A). One interleave of 30 rotations of the constant density spiral readout for a 390mm field of view (B).

Figure 2: Post-processing of the B₁⁺ map for use in MRF dictionary matching demonstrated on a phantom modeling a 22-week GA fetus. The magnitude image and B₁⁺ map are acquired (A,B), and then smoothed by fitting 6^th-order polynomial (C). The resulting difference image between the actual map and the smoothed map, which is only a 3.5% error averaged across the FOV (E). In vivo B₁⁺ maps sometimes have motion artefacts (D) that are eliminated with smoothing (F).

Figure 3: T₂ versus T₁ values for NIST and 22-week phantom. The symbol marks the average parameter value and the error bars show standard deviation of the voxels within the ROI. The ROIs are represented by color and shown in the inset: NIST phantom T₂-array (black), abdomen (blue), placenta (red), fetus (green), and brain (magenta). The measurement types are represented by symbols: reference or ground truth (x), uncorrected MRF (triangle), and B₁⁺ and slice profile corrected MRF (circle). Light gray ovals mark the individual balls in the NIST T₂-array and gray striped rectangle the relaxivity range expected in fetal tissues.

Fig 4: Example T₁ and T₂ MRF ROI maps of the brain (A, B) and placenta (C,D) of a 34-week fetus acquired during one nine second MRF scan overlaid on the proton density map.

Figure 5: Baseline brain and placenta T₁ and T₂ versus GA (A) and placenta ΔR₁ versus GA for cases where no contractions were detected during the intervening BOLD acquisition (B).

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

4072