Key Takeaways
- UNC’s Department of Obstetrics and Gynecology has earned FDA clearance for an AI‑enabled ultrasound system that estimates fetal gestational age from “blind sweep” video clips.
- The technology removes the need for expert sonographers or specialized acquisition protocols, allowing accurate pregnancy dating with minimal training.
- Invented by a multidisciplinary team (Jeffrey Stringer, MD; Ben Pokaprakarn, PhD; Juan Prieto, PhD), the innovation was licensed to Butterfly Network and incorporated into its handheld ultrasound platform.
- Prior to clearance, the system was already in research use across 12 countries, supporting maternal‑health programs in low‑resource settings.
- FDA approval now enables clinical deployment in the United States, promising to improve obstetric ultrasound access in rural, underserved areas and maternity‑care deserts.
- The milestone highlights UNC’s strength in translational research that bridges laboratory innovation and real‑world healthcare delivery.
FDA Clearance Marks a Historic Advance in AI‑Driven Obstetric Imaging
On April 17, 2026, the UNC Department of Obstetrics and Gynecology announced that its novel artificial intelligence–enabled ultrasound technology has received clearance from the U.S. Food and Drug Administration (FDA). The system, designed to estimate fetal gestational age from routine “blind sweep” ultrasound videos, represents the first FDA‑cleared tool that can deliver accurate pregnancy dating without reliance on expert image acquisition or interpretation. This regulatory milestone culminates years of interdisciplinary work aimed at democratizing essential obstetric care through machine learning.
How the AI‑Enabled Ultrasound Works
The core innovation lies in its ability to analyze standardized, non‑expert ultrasound sweeps—short video clips obtained by moving the probe across the abdomen in a standardized pattern, without needing to capture specific anatomic planes. Deep‑learning models trained on large, diverse datasets extract gestational‑age‑related features from these sweeps and produce an estimate comparable to that obtained from conventional biometric measurements (e.g., crown‑rump length, biparietal diameter). Because the algorithm handles variability in probe angle, pressure, and sweep speed, clinicians or even trained community health workers can acquire usable data with minimal instruction.
Inventors, Licensing, and Industry Partnership
The technology was conceived by a team spanning obstetrics, epidemiology, and biostatistics: Jeffrey Stringer, MD, adjunct professor in the Department of Epidemiology; Ben Pokaprakarn, PhD ’22 (biostatistics); and Juan Prieto, PhD. Their collaborative effort combined clinical insight with rigorous statistical modeling to produce a robust algorithm. UNC’s Office of Technology Transfer licensed the invention to Butterfly Network, a leader in handheld ultrasound devices. Butterfly has integrated the AI module into its existing platform, allowing users to obtain gestational‑age estimates directly from the device’s touchscreen interface.
Expanding Access to Essential Obstetric Care
By eliminating the requirement for highly trained sonographers and sophisticated imaging protocols, the AI‑enabled system opens the door to accurate pregnancy dating in settings where conventional ultrasound services are scarce, absent, or prohibitively expensive. This capability is particularly valuable for prenatal care programs that rely on timely gestational dating to schedule interventions, assess fetal growth, and manage obstetric complications. In low‑resource environments, the technology can be deployed by midwives, nurses, or community health workers after brief training, thereby extending the reach of quality obstetric assessment.
Global Research Use Prior to FDA Approval
Before securing FDA clearance, the underlying AI model was already in active research use across twelve countries, supporting maternal‑health initiatives in regions ranging from sub‑Saharan Africa to Southeast Asia. Field studies demonstrated that the system’s gestational‑age estimates fell within acceptable clinical margins of error when compared with gold‑standard measurements obtained by expert sonographers. These international validation efforts not only refined the algorithm’s robustness across diverse populations but also generated real‑world evidence that informed the FDA submission process.
Leadership Perspective: Reflections from Dr. Jeffrey Stringer
Commenting on the achievement, Dr. Jeffrey Stringer, Clarke‑Pearson Distinguished Professor of Obstetrics and Gynecology at UNC, emphasized the vision behind the work: “Our goal has been to decouple image acquisition from interpretation, making high‑quality obstetric assessment accessible far beyond traditional care settings.” He noted that the clearance validates years of iterative development, multidisciplinary collaboration, and a steadfast commitment to reducing disparities in prenatal care. Dr. Stringer highlighted that the technology exemplifies how academic innovation can translate into tangible public‑health impact when paired with industry partners capable of scaling production and distribution.
UNC’s Role in Translational Research and Healthcare Innovation
The FDA clearance underscores UNC’s prominence in translational research—transforming laboratory discoveries into clinically deployable solutions that address pressing gaps in healthcare delivery. By fostering ecosystems where clinicians, data scientists, engineers, and industry experts collaborate, UNC has cultivated a pipeline capable of moving AI‑driven medical devices from concept to market. This project serves as a model for how academic institutions can leverage their expertise to tackle systemic challenges such as maternal‑health inequities, both domestically and abroad.
Implications for Rural and Underserved Areas in the United States
Within the United States, the technology holds particular promise for rural communities and maternity‑care deserts—regions where hospitals lack obstetric ultrasound services or where patients must travel long distances for basic prenatal imaging. By enabling accurate gestational dating with a portable, low‑cost ultrasound device and minimal operator expertise, the AI system can facilitate timely prenatal visits, reduce unnecessary referrals, and improve risk stratification for conditions like preterm birth or fetal growth restriction. Health‑system administrators could incorporate the tool into mobile clinics, community health centers, or even tele‑medicine platforms, thereby expanding the geographic footprint of quality obstetric care.
Addressing Maternity‑Care Deserts and Health Equity
Maternity‑care deserts—areas with limited or no access to obstetric providers—contribute to adverse maternal and neonatal outcomes, disproportionately affecting minority and low‑income populations. The AI‑enabled ultrasound offers a practical mitigation strategy: by shifting the barrier from highly specialized sonographers to a broadly deployable AI tool, it lowers the cost and logistical complexity of obtaining essential prenatal information. When combined with targeted outreach and training programs, the technology can help close gaps in care access, promote earlier detection of complications, and support more equitable health outcomes across diverse populations.
Next Steps and Resources for Stakeholders
Stakeholders interested in learning more about the UNC Fetal Age Machine Learning Initiative can explore related coverage on Business Wire and Butterfly Network’s announcements, as well as the original story published by the UNC School of Medicine. For inquiries or collaboration opportunities, the UNC Gillings School of Global Public Health communications team is available at [email protected]. As the technology moves from clearance to widespread clinical adoption, ongoing evaluation, user training, and integration into existing prenatal‑care workflows will be critical to realizing its full potential to improve maternal and infant health worldwide.