Cleaning NDRs out of a Spamtrap Feed?

Friends, Romans, Countrymen......could I lean on you for some tips on how to clean NDRs out of my spamtrap feed? As I ramp it back up I want to make sure I'm tagging mail correctly. I may include NDRs and other types of backscatter in some of my calculations, but I definitely want to denote what it is, as accurately as I can.

Could you offer up some suggestions on things to look for? Null return path/sender header is the big thing, if every NDR was formatted properly, that would catch it all. But I'm seeing a significant amount of bounces that don't have a null sender, so I'm probably also going to resort to some sort of text string matching. So if you have a big ole list of strings or individual string suggestions, please feel free to leave them in comments.

All feedback is welcome, and thanks!