Hi Jun and Daniel-
Just chiming in a little to say that to my own eyes your first set of screenshots looked like the mask was more in the internal globus pallidus and after the shift looks better but still more like the central nucleus of the amygdala... so I'd be careful about the initial input masks. I checked FSLeyes and they don't seem to have the structure, and yes, it's in SPM's Anatomy toolbox but I'm not sure what happened during your particular transfer/transform. Manually drawing on the group average of your nonlinearly-normalized structurals shouldn't be too time-consuming in AFNI, so you may just want to do that. Naturally, only after your eyes have adjusted more to where the structure actually is based on quality work like Zaborszky's and others.
Just my $0.02, motivated in part by empathy to difficulties dealing with small and under-studied subcortical structures. Good luck!
-Sam