INI Researchers Map Sex Differences in the Brain

USC Mark and Mary Stevens Neuroimaging and Informatics Institute (INI) researchers have used machine learning to determine key sex differences in Magnetic Resonance Imaging data in a new study published online ahead of print in NeuroImage. Part of INI's Big Data for Discovery Science initiative, the study analyzed brain scans of more than 1500 healthy youths.

Researchers found that cortical thickness in two areas of the brain-the middle occipital lobes and the angular gyri-are major predictors of sex. Specifically, females demonstrate significantly thicker gray matter in these areas when multiple regions were included in the sex difference model. The team also developed a statistical learning model that accurately predicted sex based on brain morphology in more than 80 percent of participants.

"If we can establish a better understanding of sex differences in the brain, we'll be one step closer to personalized medicine," says Farshid Sepehrband, PhD, project specialist at INI and lead author of the paper. "Once we have a baseline of neuroanatomical sex differences, we can start to explore how and why some diseases-for example, autism-manifest differently in boys and girls." Understanding the anatomic differences between male and female brains could help doctors better diagnose and treat autism in the future.

The study was a collaborative effort among USC researchers, including INI's Sepehrband, Kirsten Lynch, Ryan Cabeen, Lu Zhao, Kristi Clark and Arthur W. Toga. USC faculty from the Viterbi School of Engineering's Information Sciences Institute and the Keck School of Medicine of USC's departments of pediatrics and preventive medicine also contributed to the research, along with Ivo Dinov, PhD, of the University of Michigan.

The code from the study is open-source and public via Github, where users can generate the same figures published in the paper. Sepehrband also used Plotly to generate an interactive plot of the results: hover over each plot point to see the sex difference statistics.


Source codes (Github):

Interactive plot (Plotly):