A.I. could aid research on brain trauma

Montreal is all about artificial intelligence.

We’ve heard about its impact on business and industry and how Quebec pledged in May to commit $100 million to AI over the next five years. But AI is leaving its mark on neuroscience as well and could help clarify research on concussions and brain trauma — topics that continue to give scientists headaches.

A preliminary study — conducted by the Montreal Neurological Institute and Hospital, Université de Montréal and the Ludmer Centre for Neuroinformatics — used AI software to detect the longterm effects of brain trauma on former athletes, something that could only be done in the past after the affected individuals were dead.

“It’s a very, very powerful technique,” said Sebastien Tremblay, a PhD researcher at the Montreal Neurological Institute and Hospital. “It’s definitely revolutionizing the neurosciences right now.”

The study, which was released in May and led by Université de Montréal researcher and neuropsychologist Louis de Beaumont, examined the brains of former male hockey players and football players between the ages of 51 and 75. They were split into two groups of 15; one group had concussions during their varsity years, and the other never had concussions.

After using a host of brain-imaging techniques to examine the brain tissues, researchers used AI software to crunch the data and found differences in white matter — or what Tremblay calls “the highways of the brain” — between the two groups. Tremblay said these highways channel information to different parts of the brain, which he calls cities. These cities are composed of grey matter.

“The white matter are highways that run between these cities that allow the transportation of information,” he said.

The AI software extracted the most relevant information from different imaging techniques and found white matter abnormalities in the brains of former athletes affected by concussions.

“This is very interesting,” Tremblay said. “We don’t find that many differences between concussed and unconcussed athletes in the grey matter — so in the cities, the processing units. But we do find many differences in the highways connecting them.”

Tremblay said AI software is not available in clinical settings yet, adding the findings need to be replicated using greater sample sizes.

The results correlate with previous research that found a decrease in cognitive functioning in groups of older adults concussed at a younger age. Tremblay said physicians lack the diagnostic tools to detect concussions in patients.