Quantitative analysis, image processing, and high-throughput techniques for neural imaging in C. elegans
Zhao, Charles L.
MetadataShow full item record
The use of image processing and quantitative feature extraction in the biological sciences has become increasingly prominent in recent years, as advances in technique allow the collection of increasing amount of high-quality images. With high-volume, quantitative phenotypic descriptors, it becomes possible to elucidate previously unseen aspects of the genotype-phenotype relationship, making the efficient parameterization and statistical analysis of large numbers of images more important than ever. By developing and applying image processing techniques in the model organism C. elegans, we explore the relationship between synapse-affecting genes and synaptic morphology, as well as the tracking of the real-time functional activity of neurons throughout the head ganglion on a large scale ("whole brain" imaging), seeking to develop novel methodologies of value to the whole community. By expanding a pre-existing imaging and processing pipeline to dimmer and more precise synaptic markers, we more broadly and accurately characterize the effects of already-established synaptic mutants. We then use this pipeline to perform a novel application of Quantitative Trait Loci (QTL) analysis on fluorescently-labeled synapses, a previously-infeasible study, demonstrating quantitative genetics on a subtle, different to phenotype feature, and identifying a potential QTL affecting synaptic morphology on chromosome IV. Finally, we turn our attention to function, examining the problem of monitoring many neurons in the head ganglion of C. elegans simultaneously, developing a segmentation, tracking, and data processing pipeline that requires no manual correction, and which can process volumetric neural videos far faster than previous manual approaches.We demonstrate this both by reproducing the manual curation of published videos, and by analyzing a large number of videos taken ourselves. We expect the techniques and algorithms developed to be of broad value to researchers, providing a valuable new approach to QTL analysis in previous infeasible cases, and allowing for the first time the efficient processing of "whole brain" imaging data.