Synopsis

With an increasing proportion of fetal cardiovascular MRI scans being performed at 3 Tesla, there is growing need for an accurate calibration, which characterizes relationships between MRI properties (T₁, T₂, susceptibility) and blood properties (oxygen-saturation, sO₂ and hematocrit, Hct) at 3T. Accordingly, relaxometry measurements were performed at 3T on cord blood specimens (N=89) with a broad range of hematocrits (0.09<Hct<0.82) and oxygen-saturations (7%<sO₂<100%). We also measured fetal blood susceptibility, which, to our knowledge, has never been reported. The data were effectively described by a simple, two-compartment model for blood.

Introduction

With the help of fast, magnetization-prepared pulse sequences, relaxation time and susceptibility measurements in the vessels of the late-gestation human fetus are now achievable^1-3. Using appropriate calibration data, these relaxometry and susceptibility measurements can be used to estimate hematocrit (Hct) and/or oxygen saturation (sO₂). In a recent publication, we described the relationship between fetal blood T₁ and T₂ and Hct and sO₂ at 1.5 Tesla⁴.

The prospect of a two-fold SNR boost has increased the proportion of fetal MRI scans performed at 3 Tesla⁵. With this in mind, the present study extends our previous work by characterizing the relationship between fetal blood relaxation times and Hct and sO₂ at 3T. We also include measurements of fetal blood susceptibility, which, to our knowledge, have never before been reported.

Methods

Specimen Handling: Umbilical cord blood samples were harvested from 6 human placentas following caesarean at >37 weeks gestation. The 30-50mL of blood obtained from each collection was divided into 15, 2mL specimens, which were processed (by removal/addition of plasma and nitrogen gas exposure) to provide a range of Hct and sO₂. Specimens were kept in sealed vacutainer tubes and placed in a 0.2 mM Gd DTPA doped water bath during scanning. Bath water was heated to 37°C during relaxometry, but not susceptibility measurements. Susceptibility was measured at room-temperature to avoid convection effects within the bath, which could corrupt phase subtractions. Given the limited effect of temperature on susceptibility⁶, our findings remain applicable to in-vivo conditions. Following MRI, specimen Hct and sO₂ were measured using an ABL-800 FLEX Blood Gas Analyzer.

MRI Acquisition

Relaxometry: T₁ was measured using a series of standard inversion recovery (IR) acquisitions at 9 inversion times, TI, from 24–8000 ms. T₂ relaxation times were measured at 6 echo spacings (τ₁₈₀=8.4, 12, 18, 24, 30, 36 ms) using a multi-echo spin-echo pulse sequence.

Susceptibility: To isolate the blood's field contribution from those of the gas within the tube and the background B₀ homogeneity, each blood specimen was paired with a “reference” specimen - a vacutainer tube filled to an equivalent height with the solution in the surrounding bath. Along with blood specimen phase maps (Φ_blood), phase maps of reference specimens (Φ_ref) and the empty water bath (Φ_empty) were acquired. A 3D fast-spin-echo image was also obtained for construction of a susceptibility model. The procedure for calculating susceptibility from Φ_blood, Φ_ref, and Φ_empty is outlined in Figure 1.

Data Processing

Estimating Relaxation Times and Susceptibility: T₁ was evaluated by fitting the mean signal, S, in manually drawn ROIs within each specimen tube to the exponential: S=A+B·exp(-TI/T₁). T₂ was estimated by fitting ROI signal to A·exp(-TE/T₂)+B. Susceptibility estimates were obtained by determining the bath-specimen susceptibility difference, Δχ, which most closely matched the acquired phase data (see Figure 1).

Model Fitting: Data were fit to established models representing blood as a two compartment system (plasma and erythrocytes) in the fast exchange regime.^7,8 Compartments have individual susceptibilities (χ_ery/plas) and relaxation rates (R_1,ery/plas=(T_1,ery/plas)^-1, R_2,ery/plas=(T_2,ery/plas)^-1). Relative compartment sizes are determined by Hct. Due to the paramagnetic nature of deoxyhemoglobin, erythrocyte properties relate to sO₂:

\begin{eqnarray*}\chi_{ery}&=&\chi_{ery,0}+(1-sO_2)\Delta\chi_{deo}\ , \\R_{1,ery}&=&R_{1,ery,0}+(1-sO_2)\Delta R_{1,deo}\ , \\R_{2,ery}&=&R_{2,ery,0}+(1-sO_2)\Delta R_{2,deo}\ .\\\end{eqnarray*} The model for T₂ also incorporates a plasma-erythrocyte frequency shift, $$$\Delta \omega = \omega_0 + (1-sO_2)\Delta \omega_{deo}$$$ (arising from a susceptibility difference), and a correlation time, τ, for plasma-erythrocyte spin exchange. Fits to cord blood data therefore provide estimates of: χ_ery0, χ_plas, R_2,ery,0, R_2,plas, R_1,ery,0, R_1,plas, Δχ_deo, ΔR_2,deo, ΔR_1,deo, ω₀, Δω_deo, τ.

Results

Figure 2 shows T₁, T₂, and susceptibility data for cord blood specimens. Specimen manipulations provided a range of 0.09-0.82 for Hct and 7-100% for sO₂. Data are shown alongside fitted model predictions in Figure 3. Figure 4 shows data for specimens with normal, late-gestation Hct (0.4<Hct<0.5). Estimated model parameters are summarized in Table 1.

Significant improvement in T₁ data fit quality was observed when the parameter R_1,plas was allowed to vary between subjects. This is largely due to comparatively high R_1,plas for patient #1 (circular data markers in Figures 2-4), which may result from elevated plasma protein content.

Discussion & Conclusion

In general, fitted parameter values are consistent with those reported at 1.5T⁴ and similar to published values for adult blood^8,9. Our observations suggest that the data might be better described by a more complex model that included diffusion (the effects of which are more pronounced at higher field) and accounted for plasma protein content. Nevertheless, variations in relaxation time and susceptibility were largely explained by current two-compartment models. The data and parameters we report facilitate accurate interpretation of fetal blood susceptibility and relaxometry data acquired at 3T.

Acknowledgements

References

[1] Sun, L., Macgowan, C.K., Sled, J.G., et al. Circulation 2015; 131:1313-1323.

[2] Zhu, M.Y., Milligan, N., Keating, S., et al. Am J Obstet Gynecol 2015; 214:367.e1-367.e17.

[3] Neelavalli, J., Jella, P. K., Krishnamurthy, U, et al. J Magn Reson Imaging 2013; 39:998-1006.

[4] Portnoy, S., Osmond, M., Zhu, M.Y. et al. Magn Reson Med 2016; doi: 10.1002/mrm.26231.

[5] Victoria, T., Johnson, A. M., Edgar, J. C., et al. Am J Roentgenol, 2016; 206:195-201

[6] Schenck, J.F. Med Phys 1996; 23:815-850.

[7] van Zijl, P. C., Eleff, S. M., Ulatowski, J. A., et al. Nat Med, 1998; 4:159-167.

[8] Li, W., Grgac, K., Huang, A., et al. Magn Reson Med 2015; 76:270-281.

[9] Jain, V., Abdulmalik, O., Propert, K. J., Wehrli, F.W. Magn Reson Med 2012; 68:863-867.

[10] Marques, J. & Bowtell, R. Concepts Magn Reson 2005; 25B:65-78.