Trados does not recognize text in non-editable images, including in graphs, which means that linguists cannot work on that as part of the regular translation task. This is particular problematic for MS PowerPoints as images, graphs, etc., are commonplace. 

After the translation task is complete, the translator could replace the document during the desktop publishing task but that would require re-assigning the original linguist as the desktop publisher so that they could download, translate, replace the non-editable image, and upload the document back to Trados. This is equally complicated and burdensome, and may produce unreliable results. 

What is the simplest and most efficient way to manage this?