AI algorithm estimates gestational age of fetus on par with trained sonographer
New research out of NEJM found that an artificial intelligence algorithm was able to estimate the gestational age of a fetus on par with a trained sonographer performing a fetal biometry.
The study, which was conducted in North Carolina and Zambia, collected ultrasound sweeps from 4,695 pregnant volunteers. Researchers collected blind sweep videos on both a commercial ultrasound machine and on a portable lower-cost device, the Butterfly iQ.
“We trained a neural network to estimate gestational age from the sweeps and, in three test data sets, assessed the performance of the artificial intelligence (AI) model and biometry against previously established gestational age.”
Researchers report that the “deep learning model outperformed biometry, with an overall [mean absolute error] of 3.9±0.12 days”, meanwhile the biometry’s MAE was 4.7±0.15 days.
“Our data show that AI can estimate gestational age from a series of blindly obtained ultrasound sweeps with accuracy similar to that of a trained sonographer conducting standard biometry,” researchers of the study wrote.
“This performance appears to extend to sweeps collected by untrained providers in Zambia using low-cost ultrasound devices. Whether this technology can be successfully disseminated into extant healthcare systems in low-resource settings will require further study.”
WHY IT MATTERS
Maternal and infant mortality is still a major issue worldwide. According to the World Health Organization, in 2017 about 810 women died every day from preventable causes related to pregnancy and childbirth.
Low income countries are most impacted by maternal deaths. WHO reports that in 2017 the maternal mortality rate in low-income countries was 462 per 100,000 live births, compared to 11 per 100,000 live births in high-income countries.
The study’s researchers pitch this technology as a tool to potentially help maternity care in both high- and low-income countries.
THE LARGER TREND
Butterfly Network landed FDA clearance for its ultrasound scanner that uses easy-to-manufacture semiconductor chips. Recently the company rolled out a new ultrasound platform that is able to integrate into a health system’s clinical workflows.
Other companies working in the maternal care space include Royal Philips, which launched smartphone-connected ultrasound Lumify in 2015, and smart ultrasound Exo, which nabbed $220 million in Series C funding.