Den 31/08/2007 kl. 3.01 skrev Donald McLaren:
A couple of quick questions:
(1) In the Fox et al paper, they used the residuals to do their correlations
after removing various effects. Is there any advantage or disadvantage to
using the residuals to do a straight correlation rather than using a larger
design matrix to do the analysis all at once and not dealing with the residuals?
I guess the advantage is that you can use the designmatrix as a filter, and thereby filter out e.g. motion effects from the time-series of the seed region/voxels, but remember that this also reduces the degrees of freedom.
(2) How can one save out the residuals of the entire time series in SPM5?
I recently looked at this it seems that two things need to be changed:
1. In the spm_defaults.m file set defaults.stats.maxres = inf;
2. In spm_spm.m un comment the following lines:
% %-Delete the residuals images
% %==========================================================================
% for i = 1:nSres,
% spm_unlink([spm_str_manip(VResI(i).fname,'r') '.img']);
% spm_unlink([spm_str_manip(VResI(i).fname,'r') '.hdr']);
% spm_unlink([spm_str_manip(VResI(i).fname,'r') '.mat']);
% end
This will give you as many residual images as you had original images, and it seems to me that these are really the residuals, but it would be nice if SPM authors could confirm this.
Torben E. Lund
