Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Overview 

BAOBAB stands for Biometry Automation in Obstetrics and Beyond. The World Health Organization recommends one routine ultrasound scan for all pregnant women. However, in sub-Saharan Africa, few women have access to high quality pregnancy ultrasound. The main barriers identified are the lack of affordable, robust ultrasound equipment; and a lack of trained / skilled ultrasound operators.

We aim to overcome these roadblocks in this project, undertaken in collaboration with our engineering colleagues at the Institute of Biomedical Engineering and the Bill & Melinda Gates Foundation. We investigate how recent developments in low cost ultrasound technology have meant that more affordable ultrasound equipment is now coming to the market; and how advances in real time image analysis mean that computer-assisted, partly (or eventually even fully-) automated obstetric ultrasound is becoming a promising possibility. We also aim to engage a wider field of machine learning research groups, to work on problems in this healthcare area through organising an international imaging challenge here https://a-afma.grand-challenge.org/

The A-AFMA Ultrasound Challenge (Automatic amniotic fluid measurement and analysis from ultrasound video) was featured on Computer Vision News in March 2021. Read article here

An important part of this project is to develop artificial intelligence networks for gestational age estimation. Knowing the gestational age is key to all management plans for an individual woman receiving maternity care; and essential for distinguishing pre-term from small for gestational age newborn babies. Many women, especially in low- income settings, first come into maternity care late in pregnancy, and gestational age (based on late measurement of the baby) is very inaccurate. The main problem with such late gestational age estimation is that this inherently underestimates gestation in small fetuses, and overestimates it in large fetuses. In order to overcome this we are employing the power of deep learning to use image characteristics (not measurements) - in other words the appearance of fetal structures - to enhance the accuracy of this gestational age estimation.

Clinical Research Fellow and DPhil student Elizabeth Bradburn gave a presentation: "Estimating fetal gestational age based on ultrasound image characteristics using artificial intelligence" at ISUOG 2020. (ISUOG is the leading international society of professionals in ultrasound for obstetrics and gynecology). Watch video here

Funding: Bill & Melinda Gates Foundation