In pursuit of a realistic model of the seahorse – USC Viterbi

The hippocampus is one of the most important parts of our brain. It is the hippocampus-like structure at the center of the organ that is responsible for forming our new memories.

For the first time, researchers at USC Viterbi School of Engineering will develop a highly complex computer model of the hippocampus designed to function exactly like reality. The project received a grant of nearly $ 1.2 million from the NIH Brain Research Through Advancing Innovative Neurotechnologies (BRAIN) Initiative: Theories, Models and Methods for Analysis of Complex Data from the Brain.

Dong Song, Associate Research Professor in the Department of Biomedical Engineering at USC, will lead the Hippocampus Model Project, with BME postdoctoral researcher Gene Yu (now at Duke University), David Packard Chair in Engineering Ted Berger and the professor of biomedical engineering of the Dean Vasilis Marmarelis. The research team will also include Michael Bienkowski, assistant professor of physiology and neuroscience at USC’s Keck School of Medicine.

Song was part of a team that received a $ 6 million BRAIN initiative grant in 2020 to develop polymer electrodes to monitor brain signals. He said that if his previous BRAIN initiative grant was to collect data about the brain, the latest project would draw on decades of collected brain data and knowledge to create a computer model that could mimic the function of a real seahorse. The virtual model would then be crossed with a real rodent hippocampus using machine learning to ensure its accuracy, in order to better understand the biology of the hippocampus and generate experimentally testable hypotheses about how it works.

Dong Song, Associate Research Professor in the Department of Biomedical Engineering

“The primary function of the hippocampus in higher-level mammals, such as humans, is episodic memory, especially the memory of your past experiences. It’s autobiographical memory, as opposed to learning new skills, ”Song said.

“In lower level mammals, the main function of the hippocampus is space navigation, and that is the function that I will be examining in this proposal, using rats as an animal model,” he said.

Song said that while the hippocampus is one of the best-known areas of the brain in terms of anatomy and function, there is still a lot that the research community doesn’t know about how its neurons work together to form a higher level. cognitive functions such as navigation and memory.

“For me, this is the holy grail of neuroscience – how these basic mechanisms work together to create a very complex function,” Song said.

Song and colleagues will use a machine learning framework known as the generative adversarial network, which until now has been used for image processing. The framework uses two models, which must work together to create a more precise model.

A model is the realistic “generative” model of the existing world – in this case, the hippocampus – developed from the knowledge and assumptions of neuroscience. The modeling framework would then look for the deviations between this model and the real hippocampus using a “discriminant” machine learning model.

“On the one hand, we have rats that navigate their environment and generate all kinds of signals on the spatial representation. We’ll also have a rat hippocampus model navigating a virtual environment, just like the rat does, and generating signals. We want to know if these two signals are similar enough, ”Song said.

“If there is a discrepancy, we will adjust the realistic model to better fit the data. Then we do it over and over again, iteratively, until the machine learning model can no longer distinguish between the model and the real biological system, ”Song said.

Song said such a computational modeling framework was badly needed in neuroscience, based on his consultations with leading researchers in the field. “We need a general computational framework to integrate biological knowledge, assumptions and large-scale input-output data to gain a deeper understanding of cognitive functions,” said Song Said. The project would also benefit from the expertise of Bienkowski, who would provide detailed anatomical data of the structure of the hippocampus, which would be crucial in training the realistic model.

The application of the model, Song said, would primarily be to advance our understanding of the normal functions and processes of the hippocampus – how it processes information into signals from one region of the brain to another, and how the spatial information is encoded in the activities of its neurons.

“But it can also provide a tool for understanding neurological and cognitive impairment, so if you are studying a memory impairment, you can use this platform to understand how this memory deficit is formed,” Song said.

For Song, combining biological modeling and machine learning to study the hippocampus has been a passion throughout his career. This approach was first attempted when he was a graduate student working under Marmarelis and Berger, where he modeled an individual synapse in the hippocampus.

“It was a tiny part of the neuron and it got some very interesting results,” Song said. “I’m super excited that after almost 20 years I’ve been able to extend this search for a tiny synapse to the entire hippocampus, which includes millions of neurons and billions of synapses.”