Segmentation of Developing Human Embryo in Time-lapse Microscopy

Being able to efficiently segment a developing embryo from background clutter constitutes an important step in automated monitoring of human embryonic cells. State-of-the-art automatic segmentation methods remain ill-suited to handle the complex behavior and morphological variance of non-stained embryos. By contrast, while effective, manual approaches are impractically time-consuming. In this paper, we introduce an automated approach to segment human embryo in early-stage development from a sequence of dark field microscopy images. In particular, we express segmentation as an energy minimization problem, which can be solved efficiently via graph-cuts or dynamic programming. Our experiments on twenty embryo sequences demonstrates that our method can successfully segment complex and irregular embryo structures in time-lapse microscopy (TLM) sequences.