A Machine Learning-Based Tool to Estimate Gestational Age

Ana Namburete of the University of Oxford in the United Kingdom will develop a computational tool called Autodate that identifies physical features of the fetal brain from a routine ultrasound image to automatically estimate gestational age at any stage of pregnancy. Determining accurate gestational age is important for healthy pregnancy. However, ultrasound, which is the most accurate technique, can only estimate gestational age when used during the early stages of pregnancy by a trained sonographer, who are often absent in low-income settings. To overcome these limitations, they will use existing ultrasound images of fetal brains to develop software that can automatically extract structural features. Machine learning will be used to identify links between these brain features and gestational age. The ability of Autodate to accurately estimate gestational age will be validated on newly acquired images from pregnant women in Kenya.