Behind every powerful neuron is a hardworking astrocyte — or rather, behind, in front of, over, under, and around every neuron is a network of brain cells called astrocytes.
While astrocytes are known to protect and support the brain, researchers are still puzzling over what role they play in brain diseases such as Alzheimer’s, stroke, epilepsy, and schizophrenia.
Guoqiang Yu, an assistant professor in the Bradley Department of Electrical and Computer Engineering, was awarded the National Science Foundation (NSF) Faculty Early Career Development Award to develop new computational tools to interpret and analyze astrocyte activity data.
Decoding chemical communications
New research has shown that astrocytes are more actively engaged in brain processes than anyone knew, said Yu. They “listen to and proactively regulate neurons,” explained Yu, and yet how they do this, and the impact they have on normal and pathological brains, is still unclear.
Unlike neurons, astrocytes do not generate electrical impulses. They communicate with each other and with neurons using chemical signals — specifically waves of calcium ions, which aid in the formation and function of synapses, the connections between neurons.
Although recent biotechnological advances have enabled scientists to monitor astrocyte interaction with unprecedented resolution, the immense scale and complexity of the data calls for rigorous computational modeling.
At present, researchers studying astrocyte activity must rely on a human studying time-lapse images and manually tracking regions of interest. According to Yu, this not only limits the scope of analysis, but also causes researchers to miss important information encoded in the mountains of complex, dynamic data.
A communication mystery
“There are no available models that can precisely characterize how astrocytes interact with each other and with neurons” and progress in astrocyte-based brain therapy has slowed as a result, said Yu.
Yu and his team have set out to correct this by developing the first data-driven model of astrocyte activity to automatically detect calcium signaling events, and then model and quantify them.
To meet the growing demand for analysis in action, Yu and his team will be applying the model to images of astrocyte activity in the brains of live mice.
“Any information researchers can glean about the processes taking place in a living organism will be richer and more complex than what we see in culture or brain-slice data,” said Yu.
This data-based foundation will help researchers interpret astrocyte communication, and Yu hopes it will serve as a springboard for scientists pursuing new treatments for brain disorders.
Beyond the brain
Because the project will be grappling with computationally challenging problems, Yu can see broader potential applications and plans to package the computational tools he develops for use by other researchers.
“As we construct new statistical models and develop powerful generic machine learning theory and algorithms, what we generate may prove valuable for the community of computer science.”
Yu’s team has already published several algorithmic innovations developed during preliminary studies. All source code will be available through public open-source hosting websites, so that anyone can modify and tailor the code to their specific application and need.
Outreach and education
Yu’s strategy will incorporate cutting-edge computational neuroscience problems into the engineering curriculum, improve the recruitment and retention of women and minority students, and create new opportunities for undergraduates to choose computational neuroscience as a career.