Modelling healthy and diseased cell behaviour from label-free imaging

The ability of cells to adhere, migrate and communicate with each other is integral to the physiological function of tissues. Diseased cells not only change their own cell behaviour but can also affect behaviour and function of neighbouring cells. Such effects are particularly difficult to investigate since it is difficult to discriminate diseased and healthy cells in more complex multicellular tissues with current imaging techniques.
This project is in partnership with Phasefocus, a company that manufactures microscopes with the unique capability to capture label-free cell images using Ptychography. This technology, in contrast to fluorescence imaging, has no discernible phototoxic effect on cells; it can convert the cells' inherent contrast mechanisms into high-contrast
quantitative images that can be analysed in great detail. It is ideal to investigate a large number of cellular features in long-term experiments over hours or days. To exploit the full capability of the system we will develop new algorithms that enable analysis and modelling of different parameters of cell behaviour associated with healthy and diseased cells.
The objectives will be:
1) To develop methods of distinguishing different cell types from unlabelled phasefocus images. Ptychography enables many features of the cells to be measured, which can then be used to classify the type of cell.
2) To develop methods to quantify cell behaviour. A variety of cellular characteristics can form signatures of diseased cell behaviour. Some of them can be differences in cell motility, proliferation and susceptibility towards cell death.
3) To measure the cellular responses when cells encounter different cellular environments. Cellular behaviour can differ dramatically when cells encounter changes in the biochemical and biophysical properties of their environment, which can be associated with disease (e.g. cancer, fibrosis, ageing). The project will develop methods to examine how cells behave on different types of media, and how cells interact.
Methodology: The project combines expertise in cell culture and imaging (Ballestrem and Caswell) and bio-image analysis (Cootes). The student will gain insight into novel experimental cell and tumour biological techniques as well as data analysis using computer vision and machine learning (including deep learning techniques such as Random Forestss and Convolutional Neural Networks). By closely collaborating with the company PHASEFOCUS the student will be exposed to a dynamic start-up commercial environment in which the analysis methods developed will be applied in a commercial context to produce the next generation of live cell analysis instruments.