Oxygen transport across the placenta is critically dependent on blood flow in the uterine and umbilical vessels, blood oxygen content and the diffusing capacity of the placenta.¹ Clinically, Doppler ultrasound is used to measure umbilical artery blood flow as a surrogate of blood perfusion without direct measurement of oxygen delivery.² Feasibility of BOLD MRI with maternal hyperoxygenation has been demonstrated both in animal and human pregnancies.^3,4,5 In this study, we propose a method to map the timing of oxygen transport in both healthy and pathological placentae based on BOLD signal change in response to maternal hyperoxygenation. Subjects: This IRB approved study enrolled five pregnancies. Three controls: no pathological findings in placenta and normal fetal outcome; and two pathological: one with placental infarct and velamentous cord insertion, and another with significant fetal vascular malperfusion reported of placenta post delivery. Acquisition: Studies were performed on a 3T Skyra scanner (Siemens Healthcare, Erlangen, Germany) using a combined 18-channel body and 12-channel spine receive arrays. BOLD imaging of the placenta in vivo was collected using single-shot gradient echo EPI sequence with matrix 110 x 110, 70~85 slices; in plane resolution 3 x 3 mm², slice thickness 3mm, interleaved; TR = 5-7 s, TE = 30-38 ms, FA = 90°, BW = 2.3kHz/px. Total acquisition time 30 min, with alternating maternal oxygenation protocol, designed as three consecutive 10-minute episodes: initial normoxic episode (21% O2), hyperoxic episode (15 l/min), and a final normoxic episode. Processing: N4 bias field was estimated from the averaged signal in first normoxic episode, and bias field correction was applied to all time points. Intra volume motion was corrected using non-rigid group-wise registration and for inter-volume motion correction, pairwise registration was carried out with organ specific rigid and non-rigid body transformations in Elastix software.⁵ Outlier volumes were detected based on transformation fields and temporal signal change in each voxel, and excluded. The resulting 4D data is spatially smoothed by a Gaussian kernel, and temporally smoothed by a lowpass filter. Correlation Analysis: Temporal cross-correlations between the maternal oxygen paradigm and the BOLD signal time series in each voxel were calculated. Temporal delay (τ) between the signal and paradigm was added in the correlation analysis, and assigned as the value that yielded maximum possible correlation between the two. The fact that neither maternal nor fetal hemoglobin in placentae is highly saturated,¹ allows a big dynamic range during maternal hyperoxia for oxygen to act as a tracer to visualize its regional placental delivery. In Figure 1b-l, τ maps and correlation maps as well as baseline T2* weighted images are shown for one healthy and one pathological placenta. In the healthy placenta, the τ map exhibits a pattern that agrees with placental anatomy demonstrated in Fig1a.⁶ Spheres of low delay times resemble the central cavity of the placenta cotyledon, where the incoming blood through spiral artery arrives first (red color), and then disperse evenly throughout the villous tree. The distances between centers of each region are around 3cm, which fall in the range of cotyledon sizes.⁶ Time series of the ROIs corresponding to different delay component has been plotted. In the pathological case that was reported to have extensive avascular villi, there are large placental regions that have exceeded the detection limit of our paradigm τ > 10min, (blackened region on τ map and correlation map in figure 1f) with the total volume (300cm3) equal to approximately 30% of the entire placental volume. The histogram of τ for all cases examined are plotted in Figure 2. Further a single cotyledon in healthy case was segmented, and gave a delay time of 73 +/- 16 sec. This is in good agreement with previous estimations of the transit time from the spiral artery to the uterine vein of approximately 25s.⁷ The much broader dispersions of τ in the histograms reflect heterogeneous timing of oxygen delivery across different cotyledons. Better understanding of the timing in different type of pathology may be achieved by spatial correlation between placental pathology and in vivo placenta images. Finally we note that long-range temporal fluctuations are present in the time series, which might be removed with improved motion correction strategies. We have demonstrated a method to map oxygen delivery timing in human placenta. Healthy placentae show t map that agree with normal perfusion timing in response to maternal hyperoxygenation. Pathological placentae exhibit increased dispersion of oxygen arrival across the placenta.