Chris,
I am having some trouble understanding your proposal, but if I understand correctly, then what you are proposing would be an emulation of how Flickr pictures can be migrated to Wikimedia Commons. Go to Flickr, search for CC pictures which can be reused commercially, then put the link into this tool while you are logged into Wikimedia Commons.
<
http://toolserver.org/~bryan/flickr/upload>
If you do this, you can see an existing process. This system could be adapted for NIH.
However, this may or may not be the best system to use. An alternative might be to collect the files directly from NIH and then do a batch upload. If we did that, no new software would need to be developed, it would be easier to apply the metadata, and if we are talking about protein images, we may even be able to set up a bot which either inserts the images into articles or just creates a lot of protein articles with standard text. There would be advantages to what you propose but without knowing more context it is difficult to say what is best.
Related to this - I talked with a team last Friday who want to integrate their database of transporter proteins into Wikipedia. I gave them your contact information, because I know you had already thought of this and it seemed to me like they might want to do some of this work.