Toward an experimental and computational approach to causal analysis in behaving zebrafish larvae

Saska, Daniel (2021) Toward an experimental and computational approach to causal analysis in behaving zebrafish larvae. Doctoral thesis (PhD), University of Sussex.

[img] PDF - Published Version
Download (22MB)

Abstract

Understanding brain-wide dynamics and their relation to behaviour relies on knowledge
of the interactions of the underlying functional regions in the brain. In this work, we aim
to demonstrate the applicability and limitations of Granger Causality (GC) as a measure
of directed functional connectivity in live zebrafish larvae, offering an alternative to commonly
used undirected functional connectivity measures such as correlation. In order to
acquire whole-brain datasets, we develop μSPIM: a hardware-agnostic light-sheet microscope
control and acquisition toolset which provides functionality focused on functional
imaging, providing an open-source alternative to existing light-sheet solutions limited to
developmental imaging. Further, we present an independent closed-loop virtual reality
solution which provides a exible extension to existing light-sheet or two photon microscope
setups.
In order to demonstrate the applicability of GC to calcium imaging data, we first
apply the causal analysis to simulated spiking data generated by an integrate-and-fire
model convolved with a calcium filter. We show that the directed functional connectivity
reconstructed by GC follows the structural connectivity used to simulate the underlying
network both in bi-variate and multi-variate settings. We identify a number of constraints
on the performance of the measure in form of sampling rate, recording duration and the
number of cells in the network and show that trends in the calcium data result in poor
inference which can be mitigated by filtering prior to the application of GC. Next, we
show that conditional GC on subsets of neurons can be used to infer directed connectivity
between functionally similar neuronal circuits when analysis based on all sources is not
viable due to combinatorial and computational constraints. Finally, we show that directed
connectivity inferred using GC from calcium data collected in vivo from unstimulated
zebrafish larvae displays functional characteristics described in prior research.

Item Type: Thesis (Doctoral)
Schools and Departments: School of Life Sciences > Neuroscience
Subjects: Q Science > QH Natural history > QH0301 Biology > QH0324 Methods of research. Technique. Experimental biology > QH0324.2 Data processing. Bioinformatics. General works
Q Science > QL Zoology > QL0605 Chordates. Vertebrates > QL0614 Fishes
Q Science > QP Physiology > QP0351 Neurophysiology and neuropsychology
Depositing User: Library Cataloguing
Date Deposited: 12 Nov 2021 11:05
Last Modified: 12 Nov 2021 11:05
URI: http://sro.sussex.ac.uk/id/eprint/102875

View download statistics for this item

📧 Request an update