C. elegans is a highly stereotypic model animal with (almost) fixed lineage and numbers of cells in its five stages of development. Each individual cell can be annotated, i.e. assigned a fixed ID by means of its funcion and location in a worm. Consequently, observations made at cell level in different worms at the same developmental stage can be mapped into a "reference worm".
Microscopic images can capture the locations where particular genes of interest are expressed at cell level. Assembling a large number of such expression observations for a range of genes shall yield single cell expression profiles for each larval stage. Such expression profiles reveal which gene is activated in which cell(s) at which developmental stage. This information, potentially in combination with an analysis of differences in mutants, is expected to be a key to understanding what drives C. elegans development. To achieve this goal, a large number of worms need to be segmented and annotated in microscopic images.
As manual segmentation and annotation of a single worm can take a couple of days, a high throughput study can only be realized with the help of automated image analysis. Therefore we are developing automated methods for segmentation and annotation of each cell of C. elegans in every larval stage in high-throughput microscopic images.
D. Kainmueller, F. Jug, C. Rother, G. Myers:
Active Graph Matching for Automatic Joint Segmentation and Annotation of C. Elegans
accepted for MICCAI 2014