I have the 64 GB iPad. A 32-bit float image is about 2 MB in size (2x2x2 mm resolution). This means 1 GB can hold roughly 500 different statistical parametric maps and I could fill the iPad with 30000 images in whole! If I go up in resolution to 1x1x1 mm, this number obviously needs to be divided by 8, which still leaves us with an impressing number of 3750 images.
I doubt anyone will actually have so many statistical images, so the NeuroPub app will easily store all your current and future results which you can always carry with you on your iPad or iPhone at conferences, meetings, etc.