In principle, what you suggest would be possible. In practice, the main difficulty would be aligning the slices so that when assembled, they make a reasonably correct 3D volume. That is, you don't want the 2D slices to be translated or rotated with respect to each other. Otherwise, you'd end up with a volume like a deck of cards that hasn't been "patted flat" into a nice block. AFNI doesn't really have a tool for solving this problem. One might try to use the "imreg" program to register each pair of adjacent slices, but I don't know how well that would work.
Then there is the issue of scanning the photographs. If they are on standard radiographic film, a normal flatbed scanner wont work well - you'll need a special type of scanner designed to work with such film.
Finally, AFNI itself doesn't contain many tools for structural analyses. The FSL package from Oxford does contain a tool for gray-white extraction, as do the FreeSurfer and SureFit programs (Harvard and Washington U, respectively).
bob cox
