Please have a look to:
If your are interested I can send you some code to identify outlier
subjects, at the global level, for a given contrast of interest...
Matthew Costello wrote:
> Hello all,
> I am having a problem figuring out how to detect and remove outliers from my
> data set in spm2. I have a PET study with 3 treatment groups of 17, 17, and
> 20 subjects each, all having a pre- and post- treatment scan. I believe
> this group is large enough to remove outliers, however I have not been able
> to do so.
> I used spmd to examine my data, and it provided an image with number of
> outliers per voxel. However, that is not what I am looking for, nor does it
> indicate which ones are the outliers (as far as I know). I would like to
> remove subjects based on the difference from pre- to post- treatment.
> The only method I have thought to do this would be use imcalc, create a set
> of difference images (pre-post), and look at the mean value per image and
> exclude based on that criteria. However, I am not sure if this is valid,
> nor do I know how to actually do it.
> Any advice on how to detect the outliers would be appreciated.
> Thank you,
> Matt Costello