Y. Wang

Objectives Most human in‐vivo placental imaging techniques are unable to distinguish and characterize various placental compartments, such as the intervillous space (IVS), placental vessels (PV) and placental tissue (PT), limiting their specificity. We describe a method that employs T2* and diffusion‐weighted magnetic resonance imaging (MRI) data to differentiate automatically placental compartments, quantify their oxygenation properties and identify placental lesions (PL) in vivo . We also investigate the association between placental oxygenation patterns and fetal brain oxygenation. Methods This was a prospective study conducted between 2018 and 2021 in which dual‐contrast clinical MRI data (T2* and diffusion‐weighted MRI) were acquired from patients between 20 and 38 weeks' gestation. We trained a fuzzy clustering method to analyze T2* and diffusion‐weighted MRI data and assign placental voxels to one of four clusters, based on their distinct imaging domain features. The new method divided automatically the placenta into IVS, PV, PT and PL compartments and characterized their oxygenation changes throughout pregnancy. Results A total of 27 patients were recruited, of whom five developed pregnancy complications. Total placental oxygenation level and T2* did not demonstrate a statistically significant temporal correlation with gestational age (GA) ( R 2 = 0.060, P = 0.27). In contrast, the oxygenation level reflected by T2* values in the placental IVS ( R 2 = 0.51, P = 0.0002) and PV ( R 2 = 0.76, P = 1.1 × 10 −7 ) decreased significantly with advancing GA. Oxygenation levels in the PT did not show any temporal change during pregnancy ( R 2 = 0.00044, P = 0.93). A strong spatial‐dependent correlation between PV oxygenation level and GA was observed. The strongest negative correlation between PV oxygenation and GA ( R 2 = 0.73, P = 4.5 × 10 −7 ) was found at the fetal‐vessel‐dominated region close to the chorionic plate. The location and extent of the placental abnormality were automatically delineated and quantified in the five women with clinically confirmed placental pathology. Compared to the averaged total placental oxygenation, placental IVS oxygenation level best reflected fetal brain oxygenation level during fetal development. Conclusion Based on clinically feasible dual‐MRI, our method enables accurate spatiotemporal quantification of placental compartment and fetal brain oxygenation across different GAs. This information should improve our knowledge of human placenta development and its relationship with normal and abnormal pregnancy. © 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.