Neuronal population and network analysis tools for large-scale calcium imaging datasets
Files
Main dissertation
View Supp_Video1.avi
View Supp_Video2.avi
Date
2019
DOI
Authors
Hansen, Kyle Rollins
Version
OA Version
Citation
Abstract
Recently developed large scale calcium imaging techniques allow functional analysis of hundreds to thousands of simultaneously recorded individual neurons, resulting in exceedingly large datasets. Conventional analysis methods are not scalable for large imaging datasets collected at high speed and high pixel resolution. The efforts described in this dissertation focus on the development of analysis methods designed for large datasets, along with the application of these analytic methods to derive novel conceptual insights into how neuronal circuits function in both healthy and diseased brains.
First, an image processing pipeline and a segmentation toolbox were developed and shared as an open-source software. The processing pipeline is a parallelized version of a recently published motion correction algorithm, but which improved processing speed by 10%. The segmentation toolbox is semi-automated and provides high confidence in the spatial extent of segmented cells, with the option to integrate temporal information for the segmentation.
Next, these and additionally developed methods were used to study the effect of mild traumatic brain injury (mTBI) on neuronal circuits over consecutive days. Using a newly developed signal normalization technique, we found that immediately following a blast injury event, neurons exhibited two types of changes in intracellular calcium dynamics at different time scales. One was a reduction in basal intracellular calcium levels on a time scale of minutes. The second was a reduction in the rate of transient calcium fluctuations at the sub-second time scale. Both changes recovered one hour post blast injury, suggesting different types of neuronal damage from mTBI.
Lastly, we developed a method that allowed us to observe network differences on a trial-by-trial basis with a limited number of data points. We utilized these analysis tools to study hippocampal network responses during two learning processes, trace conditioning and extinction learning. We found a similar pattern of neuronal dynamics for both learning processes, however the single-neuron identities for each process was unique.
Overall, this dissertation describes a set of image processing, segmentation, and network analysis tools for large scale calcium imaging datasets, which were applied to analyze network changes during learning and externally induced by mTBI.
Description
License
Attribution-NoDerivatives 4.0 International