Dear Luca,
For landmark-based registration Pantazis et al. (2010, http://dx.doi.org/10.1016/j.neuroimage.2009.09.027 ) should be a good reference to start with.
By the way, with regard to PET normalisation strategies, Martino et al. (2013, https://dx.doi.org/10.1007/s12149-013-0723-7 ) are more or less comparing the mentioned pipelines, and Bilgel et al. (2015, https://dx.doi.org/10.1016/j.neuroimage.2015.06.063 ) might be interesting in your case with available T1 data for a subset of subjects only (or for anyone else reading the message at a later point, as you might know these already as well).
One could also try to Dartel-register the PET images together directly (or via the Shoot toolbox) instead of relying on segmentations of those or segmentations of the associated T1 images.
Best regards
Helmut
|