Image processing

From Liebel-lab

Zebrafish Tox-Screen image processing (Rüdiger Alshut / Markus Reischl)

• Modern High Content Screening microscopes allow rapid image acquisition from 1000s of zebrafish embryos daily (for cell based screens see article below). We developed a platform for systematic chemical compound studies in large numbers (e.g. 10.000 - 100.000 ) of zebrafish including automated image analysis.
• A robust algorithm (currently implemented in MatLab) detects whether the fish is "unhappy" , dead or alive after several hours of compound exposure.
• The algorithms is orientation independent and therefore dechorionation or pre-orientation of the larvae is not necessary.
• The algorithm detects zebrafish acquired from standard multi-well plates and removes artefacts from the image.
• An automated artificial intelligence classifier "separates" the embryos.
• Currently the algorithm provides an accuracy (proper detection) of ~99%.

• Ongoing work: Detection of distinct phenotypes from highly heterogeneous images.

• If your are interested in running a set of tests or if you have questions please contact markus.reischl (at) iai.fzk.de