Computational image analysis from the transverse plane of Drosophila embryos

The constant improvement in molecular biology techniques, as well as the continuous advances in optical microscopy, has generated an unmet demand for new tools for the analysis and processing of digital images. In this paper, we present and validate a new computational approach for quantitative analysis of mRNA and protein spatial profiles extracted from transverse-plane images of Drosophila melanogaster embryos. These images are used to collect fluorescence signal intensities from protein and RNA staining at the early developmental stages. Six main image processing tasks are addressed: image preprocessing, nuclei segmentation, cytoplasm detection (apical and basal), protein/RNA quantification, detection of the dorsoventral axis, and extraction of expression profiles. This approach enables the simultaneous determination of nuclear and cytoplasmic (both basal and apical) protein and/or RNA expression. In addition, it facilitates data extraction at an unprecedent level, with a potential significant impact to understand the establishment of morphogen gradients, gene regulation mechanisms, and cell signaling in eukaryotic organism. To our knowledge, this is the first of approach this kind. Our results show that the method is fully automated and robust to a variety of confocal image settings.