Feasibility of relaxometry measurements in human fetal vessels using magnetization-prepared, motion-corrected, fast imaging techniques has recently been demonstrated [1,2]. Due to a magnetic susceptibility difference between oxy- and deoxyhemoglobin, blood T₁ and T₂ relaxation times are sensitive to the oxygen saturation, sO₂, and hematocrit, Hct [3]. Therefore, with the help of an appropriate calibration, fetal vessel relaxometry can be used for non-invasive quantification of these properties, which would of great value in the diagnosis and management of fetal hypoxia. However, due to differences between adult and fetal blood (e.g. different hemoglobin structure, 20% larger erythrocytes, 20% reduced plasma viscosity) [4], existing, adult blood derived calibrations may not be applicable to fetal blood. Accordingly, the objective of this investigation was to characterize the relationships between fetal blood relaxation times and the oxygen saturation and hematocrit. Relaxometry measurements were performed at 1.5T on umbilical cord blood samples harvested from human caesarean deliveries.

Sample Collection & Preparation: Umbilical cord blood was harvested from 6 placentas following caesarean delivery at >37 weeks' gestation. The 30-50mL volume of blood obtained from each collection was divided into multiple 1-2mL samples, which were processed (by removal/addition of separated plasma and exposure to nitrogen gas) to provide a range of Hct and sO₂. In total, 88 samples (6 collections x ~15 samples/collection) were studied.

MRI Acquisition: Samples were scanned at 37°C (using a heated water bath) on a 1.5T Avanto Syngo system (Siemens Healthcare). T₂ relaxation times were measured using a multi-echo spin echo pulse sequence with 32 echoes and 6 different echo spacings, τ₁₈₀=6.7, 12, 18, 24, 36, and 48 ms. T₁ was measured using a Modified Look Locker Inversion Recovery (MOLLI) pulse sequence with 15 inversion times, TI, from 50-5300 ms [5]. Blood samples were manually agitated prior to each scan and scan durations did not exceed 3 minutes. Immediately following MRI, sample Hct and sO₂ was measured using an ABL-800 FLEX Blood Gas Analyzer. Sample manipulations resulted in a range of 0.19-0.76 for Hct and 4-100% for sO₂. Two plasma (i.e. Hct=0) samples were also included.

Data Processing: T₂ was estimated from a mono-exponential fit of the mean spin-echo signal in manually drawn ROIs within each sample tube. T₁ was evaluated by fitting the ROI signal vs. TI to the exponential: A+B·exp(-TI/T₁*). The apparent relaxation time, T₁*, was then used to compute the “corrected” T₁:T₁=T₁*(B/A-1).

R₁=T₁^-1 data was fit to the relationship: R₁= Hct·R_1,ery + (1-Hct)·R_1,plas, where R_1,ery and R_1,plas are the erythocyte and plasma relaxation rates [6]. R_1,ery is dependent on sO₂: R_1,ery=R_1,ery,0+r'_dHb·(1-sO₂/100). A fit to R₁ data yields estimates of paramaters R_1,ery,0, R_1,plas, and r'_dHb.

R₂=T₂^-1 data was fit to the Luz-Meiboom two-site (plasma and erythrocytes) exchange model for blood [3,7], which has 6 parameters: R_2,plas, R_2,dia+R_2,oxy, R_2,deo-R_2,oxy, τ, ω_dia+ω_oxy, and ω_deo-ω_oxy. The first three parameters include the relaxation rates of plasma (R_2,plas), deoxyhemoglobin (R_2,deo), oxyhemoglobin (R_2,oxy), and other diamagnetic intracellular components (R_2,dia). The fourth parameter, τ, is the erythrocyte-plasma exchange correlation time. The final two parameters relate to the frequency shifts for oxyhemoglobin (ω_oxy), deoxyhemoglobin (ω_deoxy), and residual diamagnetic sites (ω_dia).

T₁ and T₂ data (τ₁₈₀=6.7 ms) for all cord blood samples are plotted in Figure 1 and 2, respectively. Relationships between relaxation times and blood properties are apparent: both T₁ and T₂ decreased with Hct and increased with sO₂. However, T₁ was primarily affected by Hct, while T₂ exhibited a larger dependence on sO₂.

T₁ and T₂ data along with the model fits is shown in Figures 3 and 4. Fitted model parameters are summarized in Table 1. Variations in relaxation time with blood properties were effectively explained by the models. Figure 5 shows plots, for each τ₁₈₀, of T₂ data for samples with 0.4<Hct<0.6 (the normal Hct range). Also plotted is a calibration equation of the form R₂=R_2,0+K₀(1-sO₂/100)+K₁(1-sO₂/100)², which can be employed to calculate sO₂ from a fetal vascular T₂ measurement. Coefficients R_2,0, K₀, and K₁ were calculated using the fitted Luz-Meiboom model parameters.

Simple biophysical models previously developed for adult blood were used to describe relaxation times in cord blood as a function of Hct and sO₂. MRI properties of cord blood are appropriate to describe late-gestation fetal (>35 weeks) and neonatal (<1 month old) blood. The data and fitted model parameters we report can thus be used for the interpretation of future MRI investigations of fetal and neonatal blood physiology. As the first characterization of fetal blood relaxometry at 1.5T, this work represents a key step in adaptation of non-invasive vascular oximetry techniques towards a fetal population.