Dear Gabriel,
As you have T1 images for a subset of the subjects only you would introduce a bias when relying on those images somewhere in the preprocessing pipeline for normalisation purpose and/or for creating a template. You could go like Luca has suggested, but then the MRI-derived PET template is based on those 60 subjects only and not 100. Whether the bias is relevant is another issue, but there is one.
Best
Helmut
|