Abstract

Zero-resolution conventional flow cytometry (CFC) remains the “gold standard” technology for rapid, multi-parameter phenotyping of cells in heterogeneous systems. However it relies heavily on labelling defined cellular targets with fluorescently tagged antibodies and/or reagents that, based on the resulting staining profile, allow us to partition cells in to different phenotypes, functional classes and transition states. While such directed measurements are powerful the associated reagents are expensive, add extra methodological steps to protocols, and may even induce confounding changes to cell biology particular as antibody binding can lead to the internalisation of receptors, signal transduction and alter the cells perceived identity. To this end we have been performing a high throughput, high content Imaging Flow Cytometry (IFC)-based screen to collect both directed fluorescent and label-free (bright-field and dark-field) images on a per cell basis for over 380 human and mouse CD markers under different conditions of activation. We then use the targeted fluorescent “ground truth” to uncover label-free correlates/substitutes for the associated cellular phenotypes using machine and deep learning on the transmitted bright-field and scatter images. So far we have successfully identified key immune cells involved in multi-system allergic inflammatory disease. Our comprehensive screen of label-free features that correlate with and potentially substitute for the expression of various CD markers moves us toward a library of “label-free CDs” that could be exploited for rapid “bench to bedside” diagnostic testing.