This question, I suppose, applies to all kinds of 'new' observational equipment that allow us to observe parts of nature that we weren't able to observe before. This may include the electron microscope when it was first invented. Another example may be the fairly new 'quantum microscope' which can be used to receive images of electron orbitals in atoms (or, at least, something similar).
How do we check that these instruments, when they are first designed and built, are giving us 'accurate' images such that what we observe is really what is there.
For example, with a (low power) microscope, we can use it to observe things that we can already observe with our naked eye and therefore confirm its accuracy. The same with electron microscopes on some scales. You can observe an ant with the naked eye and when an ant is put under an electron microscope, although there are now more details, you can tell its an ant (the ant's structure i.e. legs, head, antenna etc... are still visible).
Is the same thing possible when using very powerful equipment like scanning tunnelling microscopes or is some other process used all together? It seems to me that if you're inventing some observational equipment that allows you to see parts of nature never observed before, you would need to be able to check in some way that the image corresponds to something "real".