Blasi, T. et al. Nat. Commun. 7, 10256 (2016).

Teasing apart complex phenotypes in single cells at high throughput is a formidable challenge. Flow cytometry can effectively detect cells with certain features if fluorescent antibodies or dyes are available; image flow cytometry, which obtains spatial images in addition to fluorescence intensities, can find more complex phenotypes such as cell cycle phases of individual cells. Blasi et al. now show that quantitative analysis of bright- and darkfield images collected during image flow cytometry enables cell cycle measurements without any fluorescent markers. The authors extracted measurements of cell morphology, such as size, shape and granularity, from images of either fixed or live cells and applied supervised machine learning algorithms to determine cell cycle stages.