I have a proposal for a solution, though it is somewhat inelegant. How about giving the option to save an excerpt as an image instead of using OCR to convert it to text? Similar to the rectangle tool, but it won’t capture extra words at the beginning and end.
I read a lot of chemical engineering papers and textbooks and the OCR often outputs gibberish when I’m trying to excerpt something with an equation since there are a lot of special symbols and subscript and superscripts. I actually end up using the rectangle tool anyways, but it ends up capturing unwanted text before and after the desired excerpt.
Apologies if there is already a solution to this.