1416-B - A robust universal technology on reliable automated culture dishinspections

Monday, February 5, 2018
5:00 PM - 6:00 PM

The author presents a structured and modular approach for many, maybe the most, laboratory inspection tasks on culture dishes. Having multiple solutions working in the lab made it possible to find out what is common and to determine their inner basic structure. What's left were some easy to understand internal interfaces and a syntax that is leaned toward the already used notation and language of laboratory technicians and scientist. So they won't need any background in image-analysis and laboratory automation to change their manual processes towards automated ones.

With the use of closed loop imaging (CLI) capture systems from the Smart Lab Systems Group (SLabS) at Technical University Dresden, nearly all relevant properties of (micro)biological structures on culture plates can be derived. Even complex information like height profiles and reflection properties are available and used as inputs, so it is much more than simple image analysis. They are selected by the need of the analytic application and handed over in properties and dimension that fit exactly to task and that are a familiar to the user. They are mainly from category like shape, elevation profiles, colors and surface structure, as they are used in manual applications.

These results can be forwarded to either an AI or deterministic statistical identification system. The author is proving some in field use cases. This is the first time a image analysis system offers the full path from the pure sample to a useful result without any intermediate user-interaction, while it is still internally configurable. It is not just another colony-counting tool, rather that the common and complete frame for all of them, including an interface usable for automation environments.

The results will be scientifically published and are available as open source prototype technology. It can be directly adapted by industrial partners. There is an ImageJ Plugin available, as also an efficient C++ library for the framework and analytic core.

Patrick Oberthür

research associate
TU Dresden, Chair of Bioprocess Engineering
Dresden, Germany

In 2003 graduation (Abitur) Berlin; Four years training and mission duty officer at german Airforce. From 2007 to 2014 study of mathematical engineering at TU-Dresden. From 2011 to 2014 student employee as leading algoritmic developer of embedded smart cameras at Silicon Micro Sensors GmbH. Since 2012 student assistant in discrete Simulation and high-performance visualisation solution. Since 2015 research associate for image analysis systems and laboratory automation at the chair of bioprocess engineering, Technische Universität Dresden.