Hi
You might want to include the scanner as a confound in your model as was done in
Interpreting scan data acquired from multiple scanners: A study with Alzheimer's disease
NeuroImage, Volume 39, Issue 3, 1 February 2008, Pages 1180-1185
Cynthia M. Stonnington, Geoffrey Tan, Stefan Klöppel, Carlton Chu, Bogdan Draganski, Clifford R. Jack Jr., Kewei Chen, John Ashburner, Richard S.J. Frackowiak
Note that scanner scanner effects could be massive relative to any effects you are looking for, that is why you might want to change your model.
good luck
sharon.