Most of currently used high throughput-cell sorting techniques employ the analysis of low resolution data coming from the multiparametric measurements of the light peak intensities in the emission spectra of fluorochromes that are conjugated to antibodies binding to specific phenotypic cell markers. The use of more complex approaches, such as those that rely on the analysis of image-based informations, is basically limited by the highly demanding computational power that deep learning algorithms would require to quickly process the huge amount of informations confined in these dimensional data. Nao Nitta, and colleagues, however, proposed an interesting alternative.
The numerous authors of this paper presented, on Cell this innovative “intelligent image-based activated cell-sorting” (IACS), which is based on “high throughput cell microscopy, focusing and sorting”, as well as deep-learning networks, that should be able to overcome the trade-off between accuracy and speed of highly efficient – but computionally demanding – cell-sorting hardwares.