Image-based modelling of cell membrane dynamics in cell migration and cell drinking

Cell migration and the large-scale uptake of extracellular fluid during cell drinking or the sampling of antigens critically depend on a cell’s ability to change its shape. These shape changes are driven by the actin-cytoskeleton. I am going to present an overview of different modelling techniques we have employed over the years to fit mathematical models of signalling to the actin cytoskeleton and shape changes to image time series data, briefly touching on the problem of identifiability. I will conclude with introducing our recent attempts to use generative adversarial networks to create synthetic image data for training neural networks. These hold great promise to tackle difficult problems in image segmentation, serving as “silver standard” where obtaining a “gold standard” in the form of manual annotation is not feasible.

Till Bretschneider is a Professor of Systems Biology in the Department of Computer Science at Warwick University. He did his PhD at Ludwig-Maximilians University, Munich (1995-1998), developing agent-based models for understanding how three-dimensional waves of a chemical attractant orchestrate complex cell movements during multicellular development of social amoebae (Dictyostelium). As postdoc in applied mathematics (University of Bonn, 1998-2001) and fellow at the Max-Planck Institute for Biochemistry (Munich, 2001-2007) he developed new image analysis tools for investigating the dynamics of the actin cytoskeleton in cell movement. After moving to Warwick in 2007, he adopted a genuine image-based modelling approach, developing predictive models for how cells change shape during migration or fluid uptake. Using state-of-the-art lightsheet microscopy, work is now done mostly on high-resolution 3D timeseries data, employing fast GPU computing, machine learning and AI tools such as generative models.