Mothers-to-be have become used to the first glimpse of their baby via the fuzzy black and white ultrasound scan, an image that can be shown to friends and family. But it remains a luxury in many parts of the world. Now AI is being used to develop technology to bring the much-anticipated pregnancy milestone to women who are most in need of the scan’s medical checkup on a baby’s health.
A pilot project in Uganda is using AI software to power ultrasound imaging to not only scan unborn babies but also to encourage women to attend health services at an earlier stage in their pregnancies, helping to reduce stillbirths and complications.
Low- and middle-income countries tend to have very few trained specialists to carry out scans, and equipment is concentrated in urban hospitals that can be a long and costly journey from rural women’s homes.
Yet seeing women early in pregnancy can be a matter of life and death, says Dr Daniel Lukakamwa, an obstetrician and gynaecologist at Kawempe national referral hospital in Uganda’s capital, Kampala, where they are helping develop the AI software.
“It has made pregnant mothers so interested in coming for ultrasound scans,” says Lukakamwa. “People are very much willing to join the study without any apprehension. Apparently, we are getting overwhelmed.”
Late presentation is a significant issue for maternity services, says Lukakamwa. “The first trimester is very important, as far as abnormalities and maybe subsequent fevers is concerned, that lead to stillbirth,” he says.
The AI-based ScanNav FetalCheck software, made by Intelligent Ultrasound, will allow accurate dating of pregnancies without the need for specialist sonographers who take measurements of a child in the womb to determine the progress of a pregnancy. The software has been trained on a database of millions of images to “recognise” a pregnancy’s gestation.
It is one of several AI programmes assessing gestation that are in different stages of testing, and early results are encouraging, according to developers.
The technology also allows a scan to be carried out by midwives or nurses, who need only swipe an ultrasound probe over the woman’s abdomen before the programme provides data. It can also be teamed with portable devices, making possible care for women at home.
At Kawempe hospital, the aim of the trial is to create a tool to predict which pregnancies are at greatest risk of stillbirth – but staff say it is already helping them engage with women earlier.
Joness Biira, a radiographer, says: “Mothers who have delivered have made us referrals to come for the study. They talk to them, and more mothers are coming to join the research programme. They really like it; they trust our results.”
The main challenge facing staff, she says, is “maybe power failures”.
For Sarah Kyolaba, 30, who lives in Kikoni village, the technology has made her feel more in control of her second pregnancy.
“I get to know how the child is moving,” she says, “how the inner organs are developing and I get to see each and everything when I am on the scan. I am happy to see how my baby is bouncing.”
When she was pregnant with her now five-year-old, she found out at the last minute that he was too big and she would need a caesarean. “They told me I had to go for a C-section, yet I wasn’t ready for it.”
AI is also being used in the largest ever study of aspirin to prevent pre-eclampsia, taking place in Kenya, Ghana and South Africa. It will compare the effects of two different drug doses among women at high risk of developing the condition.
after newsletter promotion
An accurate gestational age is vital to the trial. Pre-eclampsia risk changes as pregnancy progresses, and aspirin’s prophylactic effect depends on it being given early.
Dr Angela Koech, an obstetrician working in rural Kenya and research scientist at Nairobi’s Aga Khan University, said accurate dating was vital.
“The biggest challenge I face is when I have a mum, usually late in pregnancy, who has developed a complication and I need to make a decision about her,” says Koech, who has led research guiding the development of AI ultrasound programmes.
“So for example, if a woman has pre-eclampsia, or hypertension in pregnancy, and she comes to me late in pregnancy, I could say, ‘let’s deliver this baby now, because your hypertension is severe’.
“But I need to know, will this baby survive? So if this woman is 30 weeks, or 32 or 34 or 36 or 38, the decision that I make will be very different.”
It would not be safe, she says, to deliver a very preterm baby at a rural facility with no neonatal unit. “When the mother tells me her gestation by last period, which she’s not sure of – if I can’t really depend on it – then it makes that decision very difficult.”
Many rural Kenyans do not see a medical professional until very late in their pregnancy, Koech said. In some places it is considered “inappropriate to reveal your pregnancy before it reveals itself”, while a visit to an antenatal clinic can cost money for travel and take many hours.
Dr Aris Papageorghiou, co-founder of Intelligent Ultrasound and clinical research director of the Oxford Maternal and Perinatal Health Institute, is aware of concerns that the technology could be seen as offering a lesser service to women in poorer countries.
“The right thing to do is to create capacity in those settings, create the right equipment, have training, have everything that we have in high-income countries. But the reality is that for the last however many decades, it has just not happened. So I think an interim solution – and it may be just an interim solution – is a good one,” he said.