Modeling Brain Activity: Network Inference and Response Simulation of Spiking Neurons

Michael Insana, University of Illinois at Urbana-Champaign

Usage Details

We are developing a large scale simulation of a network of neurons in order to emulate and predict experimentally produced action potential data for 100,000 plated rat brain neurons in vitro. The experiment, performed in the Opto Neuro Technology Lab led by Dr. Parijat Sengupta, allows for the neurons to grow and form connections over many days. Depending on the growth level of the network, phenomena such as self activation and synchronization and self sustaining of these neurons are observed. These networks of neurons are used as models to understand the mechanisms that alter neuronal network signaling efficiency post brain trauma and during neurodegenerative diseases. The networks are probed using Optogenetic tools, and measured using Multi-Electrode Electrophysiolgy, Fluorescence Correlation Spectroscopy, Calcium Imaging of Genetically Encoded Calcium Indicators, and other analytical tools. We aim to reproduce results obtained by these experiments using our model simulation. After that we will use the model to build a predictive tool to be used as a preliminary investigation resource for performing these experiments, reducing cost and time of these experiments. Our simulation models each individual neuron using Morris-Lecar model, which preserves the qualitative characteristics of the models developed by Hodgkin-Huxley, while requiring far fewer computational resources. The development considers probabilistic effects based on the spatial distribution of the neurons according to the experimental set up. For other parameters that determine detailed network characteristics of the actually experimental set up, we will be working with Dr. Boppart and his group who have developed a novel technique for using Optogenetic tools to determine neuron connections in experimental set ups like those considered in this project. Noise is introduced into the system which allows for possibility of self-activation of neurons once network connectivity has matured. Preliminary results show typical population level dynamics are captured with this model which takes the approach of an individual based simulation. In the final stage of the project we will explore the possibility of using the model in developing a control system for behavior of these neuronal network based on external stimuli.