A decade after Alan Evans and Katrin Amunts made waves in the neuroscience world with the BigBrain, unveiled in Science in 2013, they are working to take their original model to the next level.
The BigBrain is an open access high-resolution 3D digital atlas of the human brain, listed by the MIT Technology Review as one of the Top 10 Breakthroughs of 2014. The advance was born from collaboration between the researchers and their respective teams at The Neuro (Montreal Neurological Institute-Hospital) and Forschungszentrum Jülich in Germany.
The current BigBrain was produced from the brain of a deceased 65-year-old man who had no neurological or psychiatric diseases. The brain was meticulously sliced into 7,404 sections, each 20 microns thick: less than one-third the width of a human hair. To create the detailed and interactive 3D brain map, each section underwent scanning, staining and manual correction before being digitally processed. This entire process took around 10 years.
Neuroscience meets artificial intelligence
Building on the BigBrain project, the Helmholtz International BigBrain Analytics & Learning Laboratory (HIBALL) was established in 2020 through McGill’s Healthy Brains, Healthy Lives (HBHL) initiative and the Helmholtz Association of German Research Centres, with Evans and Amunts as co-Principal Investigators.
“We have some of the world’s leading experts working within the HIBALL network. HIBALL has truly unique capabilities to build upon,” Evans said, citing McGill’s strengths in computational analysis and modeling and Jülich’s strengths in data acquisition and neuroanatomy.
An international collaboration between researchers and trainees, HIBALL focuses on the connection between neuroscience and artificial intelligence (AI), aiming to enhance the 2013 BigBrain and create a highly detailed 3D model of the brain with slices just one micron thick. That’s one-thousandth of a millimetre.
“The current BigBrain is equivalent in size to 125,000 MRI volumes. Our plan to extend to one-micron resolution is a long-term goal, creating a dataset that would be 8,000 times the size of the current BigBrain,” said Evans. “This would allow us to study microscopic cellular architecture over the whole brain, a very ambitious goal that’s some years away.”
Better understanding of brain function
To address this mechanical and computational challenge, HIBALL seeks to develop new machine learning tools for pattern recognition and analysis, which will offer a better understanding of brain function and perception and facilitate the development of AI systems inspired by the brain. Before these innovations, neuroscientists had to rely on complicated and dated manual methods and microscopes to study the detailed layer structure of the brain.
“For example, student Konrad Wagstyl and colleagues used machine learning strategies to track the internal six-layer structure of the cortex using the BigBrain dataset, in a fully automated fashion, over the entire brain,” Evans explained. “Other scientists can now use this information to simulate brain network function using mathematical models. This is important work that underpins many modern approaches to understanding the brain’s functional neuroanatomy.”
Beyond its theoretical purposes, the BigBrain atlas can also be used as a template to help guide neurosurgeries: “Surgeons can superimpose the BigBrain on scans of an individual’s brain to help guide them in their electrode placements to improve Parkinson’s Disease symptoms. It’s a very practical neurosurgical application.”
Democratization of neuroscience
Evans is a champion of open science, emphasizing the value of the BigBrain’s role as a freely accessible resource.
“Since its publication, over 25,000 researchers worldwide have downloaded the BigBrain dataset for use in their own research,” he explained. The BigBrain is more than a map of the brain. It includes remote data processing tools, open project meetings, workshops and online services.
As it stands, the project’s scientific progress is shown in over 100 publications and nearly 90 datasets, allowing for a deeper understanding of the brain’s organization.
“The BigBrain is an excellent example for the democratization of neuroscience. Increasingly, we have the infrastructure and tools for researchers from low- and middle-income countries to access and process the BigBrain data remotely, instead of having to ship their own data away for analysis,” noted Evans.
Evans expressed pride in Montreal’s neuroscientific prominence, tracing it back to Dr. Wilder Penfield, one of Canada’s foremost neurosurgeons and the founder and first director of The Neuro.
“It’s amazing to be able to provide global access out of Montreal, a world capital of neuroscience,” he remarked. “McGill and The Neuro have a huge standing on the international scene. I hope more Canadians come to recognize the pivotal role played by our institutions, right here at home.”