Date of Award

January 2015

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)

Department

School of Public Health

First Advisor

Denise A. Esserman

Second Advisor

Michael J. Higley

Abstract

The reconstruction of neuronal connectivity is a very interesting and important topic in neuroscience as it helps with understanding neuronal circuit and function. With the advancement of calcium fluorescence imaging technique, we can now observe the dynamical activity of hundreds of neurons in vivo in one setting, which provides a foundation for inferring connectivity within a community or network. Poor signal-to-noise ratio and low frame rate with respect to neurons’ actual firing rate are challenges that come with calcium imaging data. Here we review several methods that can be applied to calcium imaging data, without the direct need for converting the data to spike trains which is the more traditional and popular way of connectivity analysis. We then apply generalized transfer entropy to three different sets of calcium imaging data obtained from mice visual cortex, and infer the directed functional connectivity network, in which a directed edge implies a direct causal influence by source neuron to sink neuron. The transfer entropy causal influence measure is time-dependent but requires no prior statistical assumptions on neuron firing patterns and network topology, hence model-free and applicable in face of aforementioned challenges. The performance of this measure has previously been tested on simulated data, and its performance applied to real data, as is the case in this project, is assessed using randomization. We found using properties of randomized networks compared with properties of our reconstructed network that transfer entropy was able to identify significant non-random features of the imaging data. Therefore, the inferred connectivity can provide information on the functional organization of the neuronal networks.

Comments

This is an Open Access Thesis.

Share

COinS