A new model of the human brain has been constructed by a team of IU neuroscientists and is being used to study brain diseases like Alzheimer's disease. The model provides new insight into features of the brain that haven’t previously been explored in the same way.
The model focuses on “edges,” or connections between nodes — specific regions of the brain — examining how different regions converse with one another based on their activity. Instead of simply saying regions talk to one another as past models have, the new model reveals details of the conversations.
“What our model does that’s different from anything that’s out there right now is we’re taking data and we’re approaching it from a different angle and in doing so it discloses new features of the same data,” said Richard Betzel, a professor in the College of Arts and Sciences’ Department of Psychological and Brain Sciences. “It’s exposing features of the brain that have been obscured using the existing methods.”
Betzel and Joshua Faskowitz, a graduate student who works with Betzel, became interested in the model after viewing a talk given by Yong Yeol Ahn, an associate professor in the Luddy School of Informatics, Computing, and Engineering and network scientist. While Ahn’s work focused on edge networks in general, Betzel and Faskowitz wantedto apply the ideas to the brain specifically.
A major takeaway from the model is a concept called “pervasive overlap,” which states that each region of the brain participates in multiple conversations at a time. This is a new development in neuroscience, as previous research on the topic has generally been limited to saying that correlations between parts of the brain do exist and working to track similarities in activity. The researchers’ model, however, shows that these regions often participate in more than one conversation at a time, creating an image of the brain as an overlapping edge network rather than just the sum of its individual parts.
“The way people typically think about the brain and being organized is a stylized, almost stylized view,” Betzel said. “They imagine that they can take the brain and almost carve it up into non-overlapping systems. Our model says everything is highly overlapping; basically any kind of overlap that can exist, does to some extent.”
To build the model, the research team started with recorded brain activity from 200 nodes and grouped it into pairs, which gave them 19,900 pairs and connections between those nodes. They then explored the links between the connections and estimated the conversation, which enabled them to look at similar pairs of conversations and encode them in an edge-to-edge network.
This network presents a matrix which represents a brain. By using tools from graph theory and network science to try and understand the parts of the matrix, they can then try and understand properties of the brain. Such a large amount of data creates complexities in examining it.
“One of the challenges of this new way of looking at our data is that it really explodes the number of variables we have to look at,” Faskowitz said. “The practical consequence is that you need computers with a lot of memory to even manipulate these pieces of data. Fortunately in our labs, we have really good computers.”
In terms of how brain regions interact and group, Faskowitz gave the analogy that it is similar to how humans form social networks. Though previous studies have concluded that “communities” within the brain are distinct from one another, this model shows that isn’t the case.
“In a social network you would look for friend groups,” Faskowitz said. “Your friend’s friend is more likely to be friends with you and so those are what we would call highly connected communities. Using this new concept we have, we can recover overlapping community structure, which is pretty cool because we can say for any one part of the brain we can identify that it participates in 3 or 4 or 5 different communities.”
Another key point of the model is examining how conversations develop across time. Researchers gathered data under what Betzel described as “task-free arresting conditions,” which tracks spontaneous activity. The team compared the brain activity of people watching movies versus people simply sitting in a scanner and thinking, and systematic differences were observed with different types of stimuli. Betzel said this observance of changes was a driving force for the model.
“What if instead of saying for every pair of regions they are or are not talking, what if we said more about the ebbs and flows of that conversation?” Betzel said. “What our model does is it extracts these conversations, it pulls out the dynamics, it pulls out the ebbs and flows and then it looks for other conversations that are similar.”
He also said the team is working on discovering how conversations are modulated when people are asked to do specific tasks.
Although the model is still in the early stages, looking forward, it may be able to be applied in the health care field. According to a press release, the model depicts differences in individual brain networks, which could help to classify brain disorders. Betzel said the hope is that by exposing new parts of the brain, it will give researchers the ability to develop more sensitive and accurate biomarkers.
“That’s exciting to us because the long-term goal of human neuroimaging is to identify markers of disease, to be able to detect Alzheimer’s at an earlier stage and people who are susceptible to that so we can intervene sooner; being able to understand the neurological underpinnings of, say, ASD or other kinds of developmental disorders,” Betzel said.
In addition, IU psychological and brain sciences autism researcher Dan Kennedy has begun to employ the model in terms of autism research, and IU neuroscientist Olaf Sporns is using it in dementia and memory tasks research, in collaboration with researchers at the Indiana Alzheimer’s Disease Research Center, according to the press release.
Postdoctoral associate Farnaz Zamani Esfahlani and graduate student Youngheun Jo are also on the team.
“The brain shouldn’t be understood as just a bunch of regions that do one thing. The brain is an interconnected network and brain function happens bc of all these interacting parts that form this complex network,” Faskowitz said.