Building Image Analysis Software for Live Tissue Microscopy

There is no doubt that live cell imaging will result in a myriad of novel insights into the dynamics of biological processes. Thanks to advancements in fluorescent reporter proteins and high speed microscopy systems we can now routinely generate gigantic amounts of multi- dimensional image data. However, in order to convert pixel data into knowledge, we need software tools that can accurately and reproducibly measure the properties of tissues, cells and organelles. My group focuses on the development of computational tools that enable biological discoveries from 3D time-lapse microscopy. Analysis systems for multidimensional imaging of live tissues have to address a variety of computational challenges. Extremely large data sets test the limitations of current hardware and operating systems. To measure biological parameters, image segmentation methods need to accurately segment cells and their organelles. To quantify the dynamics of cellular parameters, tracking methods need to link objects between consecutive frames. Furthermore, machine learning techniques need to be applied to translate sets of quantitative features into biological phenotypes. I will describe the development of an image analysis system for the study of cell cycle progression in live Drosophila embryos. In a case study, I will illustrate how our pipeline was instrumental in gaining new phenotypic insights into a mutation causing haploidy in Drosophila.