A gynecologist in a distant location will be able to take ultrasound images of a fetus and send them remotely to the Perinatology Institute to know the baby's state of health before birth, thanks to artificial intelligence software developed by UNAM with experts from the National Institute of Perinatology and UAM-Iztapalapa.
Fernando Arámbula Cosío, from the Institute for Research in Applied Mathematics and Systems (IIMAS), head of the "Automatic system for support in the evaluation, classification, and assignment of risks in fetuses with growth alterations", explains that this project will be ready in 2022.
"The idea is, through image analysis, to produce a diagnosis. For example, blood flow in the fetal brain, with Doppler ultrasound images that directly measure the blood flow in the brain and the software analyzes it, an estimate is made and a comparison can be made of the normal flow versus something abnormal, which the expert will then recommend," the researcher explained.
Also, a routine measurement for gynecologists to determine the size of the femur through ultrasound can be automatically analyzed by the program. The same is true for head circumference and abdomen, which should be within a standard to reveal normal growth, but if they are not, the experts will decide what to do, he said.
Arámbula Cosío and his team developed artificial intelligence algorithms that can also improve ultrasound images, several of which are difficult to interpret to improve the detection of organs, for example, the cerebellum.
The expert reported that they are currently developing an algorithm that can measure the contractions of the fetal heart, to estimate whether it is healthy, "an algorithm can be trained with normal contractions and when we record the video of a new fetus and see that the curve shrinks and expands, we could detect abnormalities concerning the normal population".
For the moment, the expert and his group have developed different algorithms that resolve each of these actions, but they still need to integrate them and put them into a single system and a single server that has remote access, as well as the capacity to process the images for analysis, said Arámbula Cosío.
What is needed to move forward?
High-performance computing. The specialist pointed out that each one of the algorithms that work with artificial intelligence requires large computing capacities to make the system work correctly, which is why it would be beneficial for them to have access to the services that will be offered through the Alianza call.
In turn, Fabián Romo Zamudio, from the General Directorate of Computing and Information and Communication Technologies (DGTIC), -who does not participate in this project-, said: soon several projects like this one will be supported with computing equipment that the private company Huawei donated at the end of 2020 to the UNAM for the creation of a new Laboratory for Teaching, Research, and Development in Artificial Intelligence and Data Science.
Romo Zamudio said: "With this infrastructure, with support and the experience of university researchers and professors, we can find alternatives for problems ranging from health issues, combating poverty, food supply, and addressing the challenges of climate change. There are many areas in which artificial intelligence can support human development. Above all, UNAM vindicates its not only technological vision but also its humanistic vision, by always having that human vision and how it can benefit society".