Joint segmentation, registration and gene expression labeling of Drosophila melanogaster imaginal discs

Date

Apr 7, 2015

Time

4:00 PM - 5:00 PM

Speaker

Archibald Pontier

Affiliation

Dept. of Cybernetics, Faculty of Electrical Engineering, Czech Technical University, Prague

Language

en

Main Topic

Biologie

Other Topics

Biologie

Host

Pavel Tomancak

Description

In this presentation, we consider the task of joint registration, segmentation and gene expression labeling. A popular method which aligns images and simultaneously estimates a simple statistical shape model was proposed by E. Learned-Miller and is known as congealing. We analyze this approach theoretically and show that the original congealing is in fact the dual - in the sense of a difference of convex functions - for a properly formulated maximum likelihood estimation task. This interpretation immediately leads to a different choice for the algorithm which is substantially simpler than the known congealing algorithm. However, our application - segmentation, registration and labeling of Drosophila melanogaster imaginal discs - demands the following two developments. We first show how to generalize the task for models in which the shape prior is formulated in terms of binary valued segmentation labeling and is related to the signal domain via a parametric appearance model to align and segment imaginal discs of Drosophila melanogaster larvae. We then proceed by enriching this model with a third label corresponding to the gene expression for each imaginal discs.

Last modified: Apr 8, 2015, 9:44:54 AM