Knowing the difference between several different types of cells is often just as critical to a cell morphology study as the location and shape of cells. In skeletal muscle, for instance, differing responses to a certain type of drug may result in drastically different changes in morphology between type I and type II fibers.
Pathologists will often stain their tissue using antibodies that are selective to a specific protein or cell type, inadvertently performing a sort of visual classification. However, straightforward computational methods to discern tissue and cell types based on the color of a stain are much more difficult to implement than it might seem to the naked eye. Multi-marker stains typically will exhibit a degree of undesirable crosstalk or co-localization, which are not easy to account for using most out-of-the-box numerical methods.
Fortunately, this need for robust cell and tissue classification has largely been satisfied by deep learning. In fact, it can often be achieved simultaneously with segmentation. Given a cell detection or segmentation model, the classification of individual cells into different categories is often a logical extension of the model and classification may be as simple as reconfiguring the output of the model for multiple output types.
Cell classification is a very commonly requested service at CytoInformatics. To see how we can help you achieve robust classification of cells in your dataset, submit a quote today.