Hi,
we are doing some VBM analyses and are now in the randomise step.
The thing is that we realised there is an outlier subject in the behavioural scores
that we can not include in the VBM analyses. Rather than re-doing all the
steps including registration of all the scans we were wondering if we could
address this by (1) giving a 0 to the input of that subject in the Group and EVs of the One-paired t test
that we will feed into randomise
or (2) shall we do fslplit on the GM_mod_merg images, remove that subject and then merge again?
Cheers
|