On Mon, 6 Jun 2005 14:26:48 +0100, Bernie Caessens
<[log in to unmask]> wrote:
>Dear spm'ers
>
>We have just encountered a for us mind boggling problem (error message) in
>setting up a second level analysis. We have 6 conditions, all contrast
>images for 8 subjects. We then enter these for 6 groups in a one-way anova.
>Our settings are like this:
>
>Global scaling NO
>Explicit masking NO
>Tresholding NO
>non-sphericity YES
>replications over repl (8)
>correlated repeated measures YES (or even no it does not matter)
>
>Then, every time again, spm_spm_ui bails out with the error message
>image orientation & voxel size are not the same
>
>We have been through the list, and understand this problem when entering
>data from unresliced images, at the 1st level, but it is our understanding
>that the con_ images should all be in the same orientation? Or is this wrong?
>However, we checked ALL of the con images with the display command, and they
>all have the exact same origin and voxel coordinates (2 2 2) !!! So they
>should be right in place no?
>
>Any hints/help would be grately appreciated.
Dear list,
I got a reply by Peter to the above problem that was most helpfull, and
hinted to what could be expected, some of our con images had a different
orientation -2 2 2 vs 2 2 2.
My question now is the following: we have ran a succesful 2nd level analysis
on the same subjects using different contrasts have 2 2 2 as orientation.
Now for our problematic analysis we only made new contrast images in which
we included the regressor of the HRF derivative (which we did not take into
account in the original analyses). Now, looking at the new con images, we
see that the orientations are very different. Is this some bug in SPM2 ? Or
did we forgot something? So to be clear, imagine 2 conditions A and B, we
have a model with A, B and a, b (derivative conv). When we did 1st level 1 0
-1 we now ran 1 1 -1 -1 and apparently this makes a huge difference in the
resulting con orientations? Any thoughts?
Kind regards,
Bernie Caessens,
Dept. of Exp. Psychology &&
Ghent Institute for Functional Magnetic Imaging (GIfMI)
Ghent University, Belgium
|