 |
 |
Dear Torsten, FEAT automatically uses the intersection of all the masks, so that you will only get results in the voxels that are present in *all* subjects. If you want another behaviour it is much more tricky to achieve, but I think this default behaviour is exactly what you want. All the best, Mark On 4 Aug 2011, at 10:41, Torsten Ruest wrote: > Hi there, > > in some of our fmri data a few top slices, and in virtually every set the complete cerebellum is missing. As the amount of these missing bits may vary from being sometimes more, sometimes less, I was wondering whether I should use an intersection mask obtained from all data present to feed it into the 2nd level analysis in order to include only tissue in the stats that's present in every subject. My understanding is that by doing so I could potentially gain power as there are less comparisons. Alternatively these differences may be neglectable and thus masking unnecessary. > > On the other hand, I am not sure how feat works: say in subject 1 there are missing superior slices, while in all the other subjects the superior bits are present. Will superior slices for subject 1 automatically excluded from mean calculation as values may be 0 or outliers (i.e. for superior slices will there be n-1 subjects going into the analysis?) ? > > Thanks very much for your input! > > Cheers, > > Torsten >