Figure 3.

While not all classes were learned from virtual H&E, (shown for 5-fold cross-validation) some show learning. Learning behavior can be seen for helper T, macrophage, epithelial, progenitor, endocrine, goblet, and fibroblast nuclei. The model’s ability to learn MxIF label information (derived from 17 stain channels) on virtual H&E (3 RGB channels) implies that there is signal present in our virtual H&E to learn some of our fine-grained nucleus subtypes.