"Deep learning is widely applicable to phenotyping embryonic development and disease"

Mutagenesis and genome editing of model organisms and subsequent screening for phenotypes is a classical approach to identify new animal models. Those are used for example for the characterization of congenital disorders. Phenotype assessment is however a very time-consuming and tedious work. Prof. Soeren Lienkamp's group published a paper, in which they show how easily deep learning methods can be applied to automate segmentation and classification of phenotypes in developing Xenopus embryos. They demonstrate how the combination of light-sheet microscopy and deep learning facilitates higher-throughput characterization of embryonic model organisms.

This work is a result of a collaboration with research groups in Germany and the USA. Co-authors of this publication are Prof. Johannes Loffing of the Institute of Anatomy and Prof. Fritjof Helmchen of the Brain Research Institute at UZH.

The open access publication can be found here.

More information to the topic can be found in the UZH press release.

Edda Kastenhuber