Transgenic zebrafish expressing fluorescent
proteins in specific tissues or organs are promising models for
studies of normal developmental processes, or perturbations of
these. However, for widespread use, reliable quantification of
the observed effects is necessary. Therefore, accurate and
automatic analysis of images obtained by fluorescent
microscopy and depicting zebrafish embryos becomes crucial.
In this paper, a segmentation approach for such images is
proposed. The segmentation is achieved by fitting a speciesspecific
2D atlas to the zebrafish data depicted in the images.
Experiments performed in a set of 50 images have provided
promising results.