Dear A Salek-Haddadi,
I bielive that it is mainly due to the reduction in RESELS which is in part due
to the smoothing step intrinsic in normalization, and also due the fact that the
brain image becomes croped in normlization as well. If you want to run subjects
unnormalized and want to get around the second part of this problem, and you
have an a prior hypothesis about which brain regions you are expecting signal
from, you can use a S.V.C on your unnormailzed data and this should return
values similar to those from a S.V.C on normalized data (assuming similar
overall smoothness of the images).
Cheers.
Alex.
> I've noticed that I get stronger activations when I normalise
> my data fMRI data. This seems to be due to the fact that the normalised
> data is smoother with a lower RESEL count.
>
> Is this simply a product of Zooming and Warping or is there
> another explanation?
>
>
>
> Yours,
>
> A Salek-Haddadi
> MRC Research Fellow
> Institute of Neurology
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|