Hey stackers,
I'd like to share this with you guys, my account rep sent me an email about the Ad Image when I asked her something regarding that and she said:
Changing the ad image doesn’t change your performance at all. It is there so we know how to categorize your campaign. Thus if you’re choosing an Image that doesn’t pertain to your category, this means you are competing with ads not related to you. It may lower the user experience as well.
So that basically means you will only find out what the best result is after the usual split testing. So go get that split testing done 
And don't forget to start with some "obvious" ones 1st and don't let the ad image mislead you, for example, the Health & Fitness ad image is with a love icon, some people might think that can be used for dating, hence it might affect your performance if you're promoting dating stuff that not related to Health & Fitness. But who knows it might convert for you, so the only way to find out is test yourself.
And feel free to use this PDF that's shared by our generous dude inversion:
http://dl.dropbox.com/u/371199/Airpush-Icons.pdf
Hope this helps some of you that still "PUSH" on Airpush.
Cheers
VERY USEFUL!!! thanks man. 
GREAT INFO! Thanks!!
Thanks for the tip dude !!!
This is what I got from my rep Sean:
"It has nothing to do with publisher/app distribution.. "
If the quote above is true I don't see split testing the images would influence the campaign...?
EDIT: Asked to be 100% sure:
"You can try it, but if there is a difference it would likely be coincidental.. It would not really be worth the time to split and test every single ad image.."
Part of it is they, like all ad networks, don't want people to submit a ton of campaigns split testing everything. More approval for them.
^^^ yeah, it makes sense. And don't forget there are lots of human errors, so I basically stick with a few ad images that somehow relate to my offer.
Good info, I usually split test images anyway, but this is great to have clarified.