Summary Project description 2nd funding period

Reliable and accurate analysis of the acquired image data is crucial in many SFB1129 projects. Given the
large volume of data, and with the aim of an unbiased analysis, the greatest possible degree of automation is
desirable. Work in the current funding period has resulted in capable image analysis software and methods
that can now successfully address most of the simpler tasks in multiple SFB projects. Careful scrutiny of failure
modes in more complex tasks has revealed two principal limitations: first, the segmentation quality that is
achievable by automated means in regimes of complex shapes, poor signal-to-noise ratio or heavy crowding;
and second, tracking errors in case of live cell image data with high object density and complex motion. In
response, in the next funding period, we propose to focus on i) making improved segmentation methods
available to the end user, ii) further improving the probabilistic tracking methods, and iii) applying as well as
evaluating these developments in multiple cooperation projects with partners from this SFB. The new
developments will exploit the enormous potential of recent advances in the field of artificial neural networks.
In summary, we will develop new methods and software that can cope with more complex and increasingly
difficult image analysis tasks in different SFB1129 projects to pave the way to significant biological findings.

Project Staff

Christian Ritter, PhD student

Steffen Wolf, PhD student