On the assumption that Dartel gives more accurate inter-subject
alignment than other approaches to spatial normalisation using SPM
(which there is accumulating evidence to support), then I would expect
the use of Dartel should give more accurate and interpretable findings.
I can not think of any specific issues relating to the registration part
of the model. The Dartel option to normalise to MNI space will combine
warping and smoothing in what I believe is a more sensible way than just
smoothing (although masking at the first level does result in some
effects that a number of people hadn't expected). I'm not sure how much
would be optimal, but more accurate alignment generally requires less
blurring in order to ensure that activations are superimposed. Without
much evidence to go on, I would probably opt for about 8mm FWHM.
I'm writing to ask your advice on the use of DARTEL in SPM8 for
normalizing functional images that I am going to be using for
It has been suggested to me to use DARTEL for normalization of
the contrast images rather than the individual scans for the
random effects analysis.
Do you think that this has any implications for PPI in terms of
extracting first eigenvariate of seed region per subject in
native space?? Do you think that it would be a good idea to
smooth the images first? Is there anything else that i should be
A very similar question has been asked before and no one has
answered as far as I can see- any help would be greatly
John Ashburner <[log in to unmask]>