While I don't think we have too much direct experience with that kind of data, we have helped some users with SWI data in the past. If you can get the phase and amplitude images separately, you can follow what some of our users have done in the literature.
[
www.sciencedirect.com]
[
onlinelibrary.wiley.com]
Alignment, masking are all done on the amplitude images and applied to the phase images. Although there are no details in those papers, I would imagine they used 3dvolreg for motion correction, 3dAutomask for masking and 3dAllineate to apply affine transformations. 3dQwarp was not yet available at the time of those papers, you can use that for nonlinear alignment instead of the ANTS package.