December 06, 2021
Mizzou Engineering students joined neuroscientists from around the world last month to exchange ideas and new discoveries about the human brain. Three undergraduate students also presented their work as first authors at Neuroscience 2021, a prestigious international conference hosted by the Society for Neuroscience.
Presenters included Pete Canfield, a junior in computer science and Matt Stroud, a senior in biomedical engineering, both from Columbia, and Greg Glickert, a senior in biological sciences from St. Louis who will start graduate studies in computer engineering next semester. They were assisted by PhD students Ben Latimer, Ziao Chen and Dan Dopp, several of whom also presented their own papers separately
All of the students work in the Neural Engineering Lab on ways to mimic complex brain cells, waves and functions using computational models.
“These students have been working on neural engineering research projects that are now mature and poised to make significant contributions,” said Satish S. Nair, professor of electrical engineering and computer science and director of the Neural Engineering Lab. “I am very impressed by their performance and work which typically does not reach such a mature stage for undergraduates. We have not had undergraduates present two international conference papers, and all three just accomplished that feat! They have been working on this for more than a year each enabling them to pursue their interest in more depth and accomplish publishable results.”
Canfield presented two papers at the conference, including work that builds on his previous development of methodology to simplify single cell design processes. This time, he incorporated machine learning to automatically determine specific parameters for cells to ensure they are as realistic as possible within the computer model. The open-source tool is available for download from CyNeuro.org.
“Neurobiologists would like to use computer models of neuron they study but are not trained to develop them,” he said. “Moreover, even modelers obtain the parameters by iterative schemes that include hand calculations, which is very costly and time intensive. Our modeling tool does all the iteration and parameter selection automatically. The idea is that biologists can provide their single neuron recordings directly as text data and our modeling tool will provide the model that fits the data, ready to run on their computer by a click of the mouse.”
Canfield also presented a biologically realistic model of the CA3, a region located in the hippocampus associated with memory processes. He used his automation tool to generate theta rhythms, or oscillations that help us learn patterns or complete routine tasks.
Stroud presented findings showing that there are different frequencies of oscillations happening in the area of the brain called M1 that controls movement of the forelimb of a rat in our Rutgers collaborator’s Lab.
“We don’t know what the oscillations are doing, specifically, but we have recorded that there are beta and gamma oscillations in the primary motor cortex,” he said. He was a co-author on two other papers from the group.
Glickert’s presentation showed a new mechanism to generate sharp-wave ripples in the CA1 region of the hippocampus. These ripples are involved in memory consolidation.
“We’re showing that researchers can stimulate a type of neuron called a chandelier to cause sharp wave ripples to occur,” he said. “The idea is to make a sandbox for researchers to test the hippocampus more easily. Chandelier cells are really small and hard to find in the brain, but we can do it easily using the model.”
He also presented a new model related to Pavlovian fear conditioning of rodents. Both of his projects were collaborations with researchers at the Queensland Brain Institute in Australia
While each project focused on small aspects of neural functions, the work helps build a better understanding of the human brain, which consists of roughly 86 billion neurons.
“Our models are simplified versions,” Canfield said. “In the real brain, it’s mind-blowing just how complex the system is. Even the best models in the world only use seven to eight compartment cells. If you want to have a very realistic single cell model you have to have something like 20,000 compartments and thousands of synaptic inputs and that’s just for one cell. When you scale that up to a whole region, you’re talking about millions of cells, and very quickly this becomes a huge problem for computational simulation.”
Glickert will continue to help unravel the mysteries of the brain after receiving a bachelor’s in biology later this month. He will return to Mizzou next semester as a master’s student and continue to work in Nair’s Neural Engineering Lab.
“There’s so much to reverse engineering the brain,” he said. “The hippocampus is only one section that I worked on; there are so many other things no one knows. Still, it felt good to present my findings after working on it for so long. I look forward to continue doing work in the lab.”