Dear Lucy,
Some ideas:
1. Brainstem is not defined properly in the AAL. You can check the Hammers and the LONI atlases, they include the brainstem at least. If you really want the pons, you may need to draw/define the region yourself.
2. I am not sure if I got your point clearly... but if what you want is to use the mean uptake value of the pons obtained from each corresponding image as the reference region for the normalization procedure, instead of the whole brain uptake as is done more frequently, then you can extract the values of the pons of each image using any software available (marsbars, FSL, or even SPM with
'spm_summarise(input_PET_image,image_with_the_pons,@mean)' ) and then use this vector of values (not 2x1 vector, it has to be one value for each corresponding image) as input in the user-defined option in the global normalization section.
Please let me know if this is not clear enough, and I can give you some extra help.