A groundbreaking new tool, which combines computer vision and artificial intelligence (AI), is on the verge of revolutionizing the way placentas are assessed at birth, according to a recent study by scientists from Northwestern Medicine and Penn State. This innovation could have far-reaching implications for enhancing neonatal and maternal care.
The research, published in the print edition of the journal Patterns on December 13 and prominently featured on its cover, introduced a computer program called PlacentaVision. This remarkable software can analyze a straightforward photograph of the placenta to detect irregularities associated with infection and neonatal sepsis, a life-threatening condition that afflicts millions of newborns globally. Northwestern contributed the largest collection of images for the study, and Goldstein was at the helm of developing and resolving issues with the algorithms.
Alison D. Gernand, the principal investigator in contact for the project, conceived the initial concept for this tool through her work in global health, particularly focusing on pregnancies where women give birth at home due to a lack of healthcare resources. “Discarding the placenta without examination is a prevalent yet frequently overlooked issue,” stated Gernand, an associate professor in the Penn State College of Health and Human Development’s Department of Nutritional Sciences. “It represents a missed opportunity to identify concerns and implement early interventions that can mitigate complications and enhance outcomes for both the mother and the baby.”
The placenta plays an essential role in the health of both the pregnant individual and the baby during gestation. However, it is often not comprehensively examined at birth, especially in regions with scarce medical resources. “This research has the potential to save lives and improve health outcomes,” said Yimu Pan, a doctoral candidate in the informatics program at the College of Information Sciences and Technology (IST) and the lead author of the study. “It could make placental examination more accessible, benefiting research and care for future pregnancies, especially for mothers and babies at a higher risk of complications.”
The scientists explained that early detection of placental infection via tools like PlacentaVision might empower clinicians to take immediate actions, such as administering antibiotics to the mother or baby and closely observing the newborn for signs of infection.
PlacentaVision is designed to be applicable across a wide range of medical demographics. “In areas with limited resources – where hospitals lack pathology labs or specialists – this tool could assist doctors in rapidly identifying issues like infections from a placenta,” Pan noted. “In well-equipped hospitals, the tool may eventually help doctors determine which placentas require further, detailed examination, streamlining the process and ensuring that the most critical cases are prioritized.”
“Before such a tool can be globally implemented, the core technical challenges we faced were to make the model adaptable enough to handle diverse placenta-related diagnoses and to guarantee that the tool is resilient enough to cope with various delivery circumstances, including variations in lighting, imaging quality, and clinical settings,” said James Z. Wang, a distinguished professor in the College of IST at Penn State and one of the principal investigators. “Our AI tool must maintain accuracy even when many training images originate from a well-equipped urban hospital. Ensuring that PlacentaVision can handle a broad spectrum of real-world conditions was of utmost importance.”
The researchers employed cross-modal contrastive learning, an AI technique for correlating and understanding the relationship between different data types – in this instance, visual (images) and textual (pathological reports) – to train a computer program to analyze placental pictures. They amassed a large and diverse dataset of placental images and pathological reports spanning 12 years, studied how these images correlate with health outcomes, and constructed a model capable of making predictions based on new images. The team also devised various image modification strategies to mimic different photo-taking conditions, enabling proper evaluation of the model’s robustness.
The outcome was PlacentaCLIP+, a powerful machine-learning model that can analyze placental photos to detect health risks with a high degree of accuracy. It has been cross-nationally validated to ensure consistent performance across different populations.
According to the researchers, PlacentaVision is engineered to be user-friendly, potentially operating via a smartphone app or integrated into medical record software, allowing doctors to obtain rapid answers after delivery.
“Our subsequent steps involve developing a user-friendly mobile app that can be utilized by medical professionals with minimal training in clinics or hospitals with limited resources,” Pan said. “The user-friendly app would enable doctors and nurses to photograph placentas and receive immediate feedback, thereby enhancing care.”
The researchers intend to make the tool even more intelligent by incorporating additional placental characteristics and clinical data to refine predictions while also contributing to research on long-term health. They will also test the tool in diverse hospitals to ensure its functionality in various settings.
“This tool has the potential to revolutionize the post-birth examination of placentas, especially in regions where such examinations are seldom conducted,” Gernand said. “This innovation holds the promise of greater accessibility in both low- and high-resource environments. With further refinement, it has the capacity to transform neonatal and maternal care by facilitating early, personalized interventions that avert severe health consequences and improve the lives of mothers and infants worldwide.”
This research was funded by the National Institutes of Health National Institute of Biomedical Imaging and Bioengineering (grant R01EB030130). The team utilized supercomputing resources from the National Science Foundation-funded Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program.
Related topics
- Short Daily Reflection Sessions Help Health Care Workers Combat Burnout
- Study Identifies Structural and Functional Brain Changes in Opioid Use Disorder
- Study Highlights Ableism as a Barrier to Mental Health Care for Disabled Individuals
