Artificial intelligence tool reveals sex differences in brain structure

Artificial intelligence (AI) computer programs processing MRI scans reveal differences in the organization of men's and women's brains at the cellular level, a new study shows. These differences were found in white matter, tissue located primarily in the inner layer of the human brain that facilitates communication between regions.

Men and women are known to suffer differently from multiple sclerosis, autism spectrum disorder, migraines and other brain problems, and to exhibit different symptoms. A detailed understanding of how biological sex affects the brain is seen as a way to improve diagnostic tools and treatments. However, although the size, shape and weight of the brain have been studied, researchers have only a partial understanding of its structure at the cellular level.

The new study, led by researchers at NYU Langone Health, used an AI technique called machine learning to analyze thousands of MRI scans of the brains of 471 men and 560 women. The results showed that computer programs could accurately distinguish between male and female brains, revealing structural and complex patterns that were invisible to the human eye.

The results were confirmed by three different AI models designed to determine biological sex, using their relative strengths in either focusing on small areas of white matter or analyzing connections across large brain regions.

"Our findings provide a clearer understanding of the structure of the living human brain, which may offer new insights into how many psychiatric and neurological disorders develop and why they may present differently in men and women," said the study's lead author. And neuroradiologist Yvonne Luey, MD.

Luy, professor and vice chair of research in the department of radiology at NYU Grossman School of Medicine, notes that previous studies of brain microstructure have relied primarily on animal models and human tissue samples. Additionally, the validity of some of these past findings has been called into question by the use of statistical analyzes of “hand-drawn” regions of interest, which required researchers to make many subjective decisions about the shape, size, and location of the selected regions. Such elections could potentially distort the results, Lui says.

The new study avoided this problem by using machine learning to analyze entire groups of images without pointing the computer to a specific location, helping eliminate human bias, the authors note.

For the study, the team began by providing the AI programs with existing data from sample MRI brain scans of healthy men and women, also specifying the biological sex of each scan. Because these models were designed to use sophisticated statistical and mathematical techniques to become "smarter" over time as data accumulated, they eventually "learned" to distinguish biological sex on their own. It's important to note that the programs were prohibited from using overall brain size and shape for their determinations, Lui says.

According to the results, all models correctly identified gender from scans in 92% - 98% of cases. Several features particularly helped the machines reach their conclusions, including how easily and in what direction water could move through brain tissue.

“These results highlight the importance of diversity when studying diseases that originate in the human brain,” said study co-author Junbo Chen, MS, a doctoral student at the NYU Tandon School of Engineering.

“If, as has been the case historically, men are used as the standard model for various disorders, researchers may miss critical insights,” added study co-author Vara Lakshmi Bayanagari, MS, a graduate research student at the NYU Tandon School of Engineering.

Bayanagari cautions that while AI tools could report differences in brain cell organization, they could not reveal which sex was more prone to which features. She adds that the study classified gender based on genetic information and only included MRI scans of cisgender men and women.

The team plans to further study the development of sex differences in brain structure over time to better understand the role of environmental, hormonal and social factors in these changes, the authors said.

The work was published in the magazine Scientific Reports.