That's a pretty good question. I think that before we try to figure out how to do it, we should ask what a 3D histogram would even look like. The problem with 3D data is that you end up needing four dimensions to visualize it. I think the practical solution would be to reduce the problem to 2D slices, but 128 slices is a lot to deal with.
Alternatively to a stem/bar histogram, you could represent frequency with an intensity map. That would reduce the dimensions required back to 3. Still, what's the most context-appropriate method to represent a volumetric image but as slices?
