Dorsal closure in numbers: quantification of epithelial cell oscillations using machine learning

Dorsal closure in Drosophila melanogaster embryos is a key model system for cell sheet morphogenesis and wound healing. Multiple sub-systems are involved in the mechanical closing of the dorsal opening. Understanding system dynamics, regulation and causal relations requires a quantitative understanding of the mesoscopic mechanical and dynamic properties of this “active soft material”. Individual cells in the amnioserosa, a one-cell-thick sheet of epithelial cells filling the dorsal opening, show sustained oscillations of apical cell area. These oscillations exhibit large variations from cell to cell and during the course of closure. Past studies of epithelial dynamics were restricted to semi-manual segmentation of cell shapes and thus suffered from relatively low statistics. We present a novel analysis pipeline, based on a convolutional neural network (machine learning), that allows an automated and robust segmentation of large numbers of video recordings. We further employ statistical approaches to analyze spatial-temporal dynamics and quantify embryo-to-embryo variability. We observe emerging long-range dynamical patterns providing clues about possible communication mechanisms between cells.