by Nikki Galovic, CSIRO

MRI scans showing anatomical connections in the developing brain. Credit: The Developing Human Connectome Project

When Briony and her husband embarked on their journey to start a family, they didn't think artificial intelligence (AI) would play a role.

For years they longed for the pitter-patter of tiny feet, but it was a dream that traditional methods couldn't fulfill. They turned to a fertility specialist and started in-vitro fertilization (IVF).

While IVF is common—of all the babies born in Australia, 1 in 18 will have been conceived through IVF—success rates vary. Less than half conceive on their first cycle of IVF and many people go through multiple cycles without success.

It wasn't an easy road for Briony and her partner. But after two rounds of IVF and two embryo transfers, Grace came along.

With one successful IVF childbirth, the couple decided to try for a second. Once again, they went through egg retrieval, this time ending up with three healthy embryos. It was at this point they were presented with an opportunity to be part of an AI-assisted embryo selection trial.

After some initial apprehension, Briony embraced the chance to help the future of IVF and other people like her get pregnant. The first transfer from this round wasn't successful. But the second led to the birth of their son Rory. And they learned Rory was selected by AI.

AI's keen eye: Seeing what we might miss

Briony's story highlights the potential of AI in fertility treatments from embryo selection and beyond.

IVF gives many people the chance to have children that they otherwise wouldn't be able to. But it can be cost-prohibitive, and the emotional strain and pressure can be incredibly taxing.

Traditional fertility methods rely on visual assessment by highly trained embryologists. But AI can analyze huge amounts of data—like time-lapse footage from under the microscope—and identify subtle patterns and trends that the human eye might miss. And it can do it in about a tenth of the time it takes an embryologist to do the same. A combined approach of AI and experts can give a more comprehensive picture and lead to a higher chance of successful pregnancies.

And it's not just used for embryo selection. AI algorithms are also being developed to help identify healthy sperm in severely infertile men.

One new AI algorithm developed by an Australian company found healthy sperm in men who have non-obstructive azoospermia. An infertility condition, azoospermia affects roughly 5% of couples seeking fertility treatment.

The AI is given a large dataset made up of thousands of microscopic photographs of droplets of sperm, complete with other cells and tissues. The sperm cells are highlighted, so the AI can learn and remember which cells to look for.

When put to the test, the AI was able to locate sperm in less than a thousandth of the time taken by humans. It also found more sperm overall in comparison to a human embryologist.

AI in fertility treatments is still a new area but it is showing promise of being a valuable tool to augment the expertise and experience of embryologists. It holds the potential to improve success rates and ultimately help more people realize their dreams of having a baby.

AI and the well-being of premature babies

AI is also making strides in supporting the development of premature babies.

Dr. Jessica Bugeja and her team at our Australian e-Health Research Center are exploring whether they can predict neurodevelopmental or cognitive challenges from premature babies' brain scans.

They're using AI to analyze MRI scans of premature babies. MRIs provide detailed pictures of the brain, allowing researchers to track development. Jessica and the team scan infants early (in the first weeks of life) and again at various stages over the next six years.

The scans identify clues about potential development issues, such as difficulties with learning or movement. They leverage AI to explore finer details beyond what the human eye can perceive. This information is then compared with data collected from specialists like physios and neurologists to create a comprehensive picture of each baby's development.

The ultimate goal is to predict potential issues like ADHD or cerebral palsy much earlier. Early detection means earlier intervention which can lead to better outcomes. This is just one way AI is being used in medical imaging.

Everyday AI: Exploring AI's impact

While the full results of Jessica's team's research are still to come, the initial findings are promising. As research progresses and data sets grow, AI's accuracy and potential applications for babies will only continue to evolve.

Provided by CSIRO