Reanimating the past: From historical collections of the placenta and uterus to modern imaging, machine learning, and...

The placenta and uterus are essential to healthy human pregnancy, yet their complex adaptation and function remain difficult to study, particularly during the second trimester. Historical collections of placenta in situ specimens, including the Boyd, Dixon, and Carnegie Collections, provide rare access to placental and uterine anatomy across developmental stages and pathological conditions. Due to advancements in obstetric care, the preparation and extent of specimens in these collections is uncommon nowadays. They therefore offer a unique window into pregnancy that contemporary imaging cannot fully reproduce. This review article briefly summarizes these historical collections and examines how they can be integrated with modern imaging, machine learning, and multiscale modeling to extend their research utility. Advances in computational methods support the digitization and 3D reconstruction of 2D histological sections, helping to mitigate issues such as tissue deformation, staining variability, and missing sections. In addition, machine learning approaches enable automated segmentation, facilitating more detailed characterization of utero-placental structures and associated physiological functions. Image-based multiscale modeling can link microscale features, such as villous capillary architecture, to organ-scale responses including global perfusion and oxygen transport, while enabling efficient computation when enhanced with physics-informed neural networks or neural operators. These methods preserve the scientific value of legacy specimens while enabling new avenues for studying pregnancy complications such as preeclampsia and fetal growth restriction. By combining legacy resources with contemporary analytical techniques in an interdisciplinary framework, a more thorough understanding of placental biology and maternal-fetal health can be achieved that may inform future research and clinical applications.