If you want to do a group ("2nd level") analysis voxel-wise, then no. But if you just care about ROI-wise response then yes.  

Not sure I understand what you mean by 'contrast value'. But you can use Marsbar to extract % signal change or beta-weights, and then import these values into SPSS. Then you can do correlations, group comparisons etc.

Yes of course, you can extract the mean signal change using rex tool and import your data in spss!!