Researchers from the UNC Institute for Global Health and Infectious Diseases and the University of Zambia collaborated on a study showing that artificial intelligence-enabled ultrasound devices can be used to provide accurate gestational age estimates. The device may revolutionize pregnancy care in areas where high-end ultrasound machines are not available.
The study, which was published in the Journal of the American Medical Association, used data collected from 400 individuals with first-trimester pregnancies in Lusaka, Zambia, and Chapel Hill. The study was funded by the Bill & Melinda Gates Foundation, and the device — which is portable and battery-operated — uses an AI tool integrated with the Butterfly iQ+ handheld ultrasound device.
Dr. Jeff Stringer, a professor at the UNC School of Medicine and lead author of the study, came up with the idea for this project while working as an obstetrician in Lusaka.
“[Stringer] is the one who wanted to use AI technology to bridge the gap between experts and non-experts in limited resource settings like middle-income countries,” Juan Prieto, who helped create the AI model, said.
Teeranan “Ben” Pokaprakarn, who also helped to develop the AI model, said that in low- to middle-income countries like Zambia, high-end ultrasound machines and trained clinicians may not be readily available. Therefore, they faced a challenge to create a device that was not only accessible, but also one that can be operated by novice users.
“We think if we do it this way, where we make the ultrasound procedure simple enough that a midwife can do it with AI technology, and we can show that it's equivalent to the standard of care approach, then I think it's a great use of AI technology,” Pokaprakarn said.
Prieto said in a typical ultrasound examination, a trained sonographer will acquire the perfect snapshot of the head and the abdomen of the fetus and then perform measurements of those fetal structures. However, without training, this can be difficult to do.
Using the AI device, healthcare workers with minimal training can instead perform a blind sweep — “blind” meaning the user swipes across the abdomen but is not able to see typical ultrasound images. Instead, the AI model interprets videos from the sweeps, synthesizes the information and provides an estimate for gestational age.
Savannah O’Brien, president of the student organization UNC Saving Mothers, said that ultrasounds are important for the medical safety of the baby and the mother. O’Brien said information collected through ultrasounds can be instrumental in making a plan for the birth and identifying any abnormalities.
