Well, there is no absolute way other than knowing the history of the dataset. You could compare dimensions and coordinate locations with a template or atlas. Still none of this guarantees a particular space. If you are looking at a whole brain dataset, you could check if the alignment is better with some MNI template versus something else, but hard to say what the "something else" is. If you are looking at a mask region, then you might look for overlap with that region in an atlas, but I am not sure what the point would be, if you can just get the region again. If the dataset is a cluster or set of clusters, then I doubt there are any absolute solutions.