Can ControlNet be used to do virtual staining of hisopathological (H&E) images?

[chokevin8]

I was wondering if ControlNet could be used to virtually stain an unstained image for my research purposes. For those not aware of how unstained and stained H&E images look like, they look like [this[(https://www.google.com/search?sxsrf=AJOqlzWGwEsKcZEGMowYq7gSsUmbJNS3VQ:1675370125229&q=unstained+to+stained+h%26e+image&tbm=isch&sa=X&ved=2ahUKEwj1ibmJ2Pf8AhUTUzUKHRiHCwcQ0pQJegQIDxAB&biw=1873&bih=969&dpr=1#imgrc=x3HbfcorYe-2yM). The unstained and stained images would have the same structure & texture (essentially the same segmentation mask) but different color obviously. I was wondering if with enough training images, the model could accurately stain the unstained images to make them look like real H&E stained images.

Would appreciate any input, thanks a lot in advance!