MBI PhD Qualifying Exam
Time: 1230pm
Date: 26 Feb 2019, Tuesday
Venue: MBI, Level 5 Meeting Room.
Supervisor: Dr Yusuke Toyama
Development of deep neural network model for inferring in situ tissue tensions
by Murat Shagirov, Toyama’s Group
Abstract
During embryogenesis, a simple egg is shaped into a complex three-dimensional organism in a process called morphogenesis. Understanding morphogenesis requires knowledge of how forces that drive morphogenesis are distributed within the tissue. Although some approaches have been developed for inferring in situ forces in native tissue there are some limitations. Major limitations of current methods lie in their inability to deal with dynamics and complex cell shapes. Here, I propose to develop a method which is a complement to current ones—a deep learning-based method to infer tissue tensions in situ from confocal movies. I use amnioserosa tissue of Drosophila embryo, which displays complex cell shape and dynamics, as a model system. To achieve this proposal, I aim to 1) develop an image segmentation tool to quantify the shape of the cell from a video, 2) develop a deep neural network model to predict a tension at cell junctions from both a static image and an associated laser ablation data, and 3) combine the above-mentioned methods to predict the tension distribution of an unperturbed tissue. In the future, I aim to make this model into a general tool for mapping tissue tensions in other tissues. More detailed preliminary results and future plans will be discussed in the presentation.
**Please note the examination following the seminar is closed-door**
