AI helps sonographers detect fetal abnormalities faster during 20-week pregnancy scan

Artificial intelligence (AI) could help sonographers identify any abnormalities at the 20-week pregnancy screening scan almost twice as quickly, without reducing the accuracy or reliability of diagnoses, a new study has shown.

This will help improve patient care by allowing sonographers to focus on other aspects of the scan, such as communicating with parents or spending more time looking at any areas of concern.

The trial is the first of its kind to use AI for the 20-week pregnancy scan on real patients, and is led by King’s College London and Guy’s and St Thomas’ NHS Foundation Trust and funded by the National Institute for Health and Care Research (NIHR).

Results from the trial, published today in NEJM AI, found that AI-assisted 20-week scans were significantly shorter than standard scans, reducing the scan length by over 40%.

For many expectant parents, the 20-week pregnancy screening can be a worrying experience as sonographers are looking for signs of 11 different conditions in a baby, such as heart defects or spina bifida. However, the extra time freed up by using the AI tool allows the sonographer more time to focus on the parents, which may help to ease anxieties.

Dr Thomas Day, lead author of the study, who is a Clinical Research Fellow at King’s College London and a Consultant Paediatric and Foetal Cardiologist at Evelina London Children’s Hospital, part of Guy’s and St Thomas’, said: “Understandably this 20-week scan can be a nerve-wracking time for parents, as they’re finding out the health of their unborn child. Our research has shown that AI-assisted scans are accurate, reliable and more efficient. We hope that using AI in these scans will free up precious time for sonographers to focus on patient care, making the experience more comfortable and reassuring for parents.”

The trial featured 78 pregnant participants and 58 sonographers. Each pregnant participant was scanned twice, once using the AI-assisted scanner and once without the use of AI.

The study showed that the AI-assisted scans – which automatically took several thousand snapshots of each foetal measurement compared to three taken by a human sonographer – were more reliable compared to manual measurements. This may improve the accuracy of assessing foetal growth and health, helping medical professionals make decisions earlier that could improve health outcomes for the baby.

The AI-tool alters the way in which the scan is performed, as sonographers no longer need to pause, save images, or measure during the scan, resulting in fewer interruptions and making the process smoother.

Ashleigh Louison, a 36-year-old senior operations manager from north-west London, was one of the participants in the trial at St Thomas’ Hospital. During her pregnancy, her son Lennox was diagnosed with heart disease. This early diagnosis allowed doctors at St Thomas’ and Evelina London Children’s Hospital to plan immediate medical care for Lennox when he was born, including lifesaving surgery within two weeks of his birth.

She said: “Receiving an early diagnosis for Lennox was really important, as it meant we could properly plan the road ahead. We immediately knew that he would likely need open heart surgery and that we would be staying in hospital for a few weeks after his birth. This gave us the chance to physically and mentally prepare for what was coming.

“I am so glad to have participated in this trial as I want to support anything that can help save children’s lives through faster and earlier diagnoses of conditions. I know that some conditions can be hard to spot, and so I’m excited at the prospect of using new technology that can help address this. If my participation in this trial ends up helping even just one family, then I’m all for it.”

The AI tool is now being rolled out more widely through a company called Fraiya, which is a University-NHS spinout company from King’s College London, Guy’s and St Thomas’ and King’s College Hospital. The company’s goal is to create AI-powered tools for pregnancy ultrasound that improve medical diagnoses, help healthcare professionals, and enhance care for patients.

Fraiya has been developed as part of the MedTech Venture Builder programme at The London Institute for Healthcare Engineering (LIHE).

Professor Reza Razavi, senior author of the study and Professor of Paediatric Cardiovascular Science at King’s College London and a Paediatric Cardiologist at Guy’s and St Thomas’, said: “It is vital that congenital problems in unborn babies are picked up in pregnancy to help give them the best possible outcomes after birth. Our study shows that combining cutting-edge AI with human expertise could make these scans much quicker and easier to do, whilst maintaining the accuracy and reliability of the scans. This is a great way of improving staff experience whilst also improving the care of mothers and babies. AI can support the scanning clinicians to work at the top of their game so every hospital across the country can really deliver great care.”

Anything that improves the experience of expectant mothers, gives them reassurance and tailors the care they are offered by healthcare professionals can only be a good thing.

The use of AI in healthcare has huge potential to impact patient care while saving time and money. This research is another great example of how NIHR works for the health and wealth of the nation.”

Professor Mike Lewis, NIHR Scientific Director