Annotations are linked by default, meaning that if you annotate the image, it will try and create the same annotation on the generated text. This is typically for highlights, redactions, etc.

In order for this to work though, the generated text needs to have come from OCRing the image page so that the there are valid image locations that can map where on the image the annotation is being made to the correct text.

You can try re-OCRing the document and then see if annotations can now be linked.