Dear All (Alex, Krish - thanks for your replies),
I should have mentioned that I had already reset the
origin of the mask images using HDR edit in SPM. So
the mask matched the T1/EPI template in SPM. When I look
at mask.img used for the analysis it agrees pretty well
with the mask I generated (there's a hole in the inferior
frontal lobe region corresponding to the signal dropout in
the actual EPI data).
Perhaps the most confusing thing, that I neglected to
mention in my 1st email, is that if I use the SVC function
in the results window and mask the activation map (obtained
with "un-masked" whole brain EPI data) with
my brain mask, I get loads of highly significant (corrected
p <0.0000) and large blobs. The reason I want to mask at
an earlier stage in the analysis, is to get the actual
activation map to superimpose over my average anatomical
brain - and the SVC function doesn't give you this.
Are the two processes I have described equivalent? I.e.
is masking the data that goes into the GLM the same
as masking "whole brain" data at the SVC stage?
My guess is that I'm doing something pretty stupid (it
wouldn't be a first), but I'm at a loss as to what
it is.
Can anyone help?
Cheers,
Jon.
_____________________________________________________
Jonathan Brooks Ph.D. (Research Fellow)
Magnetic Resonance and Image Analysis Research Centre
University of Liverpool, Pembroke Place, L69 3BX, UK
tel: +44 151 794 5629 fax: +44 151 794 5635
On Fri, 22 Jun 2001, Leff, Alexander wrote:
>Dear Jon,
>
>The simplest (and easiest to sort out) problem is that you have not reset
>your origin on the self- made mask. Analyze will write out this image with
>no associated mat file and spm99 will chose the central voxel as the orign.
>As this will not match up with you functional data (whose origin will be at
>the ac point if you apatially normalized your activation scans), your blobs
>will not fall inside your mask and you will get zilch. Use the reorientate
>image function in spm99. Display the average brain you used to create the
>mask, note down it's origin. Display your mask image and move it in the x, y
>and z plane until it has the same origin point; press reorientate image and
>the mask will be saved with a mat file. It should then work for your svc.
>
>Cheers.
>
>Alex.
>
>-----Original Message-----
>From: Jon Brooks
>To: [log in to unmask]
>Sent: 6/22/01 3:01 PM
>Subject: problems using a brain mask
>
>Dear All,
>
>In response to reviewers comments on a paper
>we just submitted, I have been asked to reanalyse
>some fMRI data using a small volume correction.
>
>Rather than pick spheres or define cubes I want
>to use a mask determined from our average brain
>(made from the anatomicals for each subject in the
>study). So using Analyze I have created a mask by
>using "image edit" to block out the relevant regions.
>
>The selected regions are set to 255 and everything else
>to 0. I then smoothed the mask in SPM99 using a 3mm FWHM
>kernel. Because I want to see the effect of the masking on
>the resultant SPMs I followed a suggestion in an earlier
>thread and applied the mask to the first time point
>of the fMRI data. This should restrict the analysis to
>a volume corresponding to the mask.
>
>Using a standard fMRI design matrix I ran the appropriate
>analysis and went to look at the results. I got zilch!
>With the unrestricted analysis and a corrected p(0.05) I
>got lots of blobs from a sample volume of 630 resels, in
>my restricted analysis only 0.1 resels are included in
>the analysis. I'm really out of my depth here as to the
>reason why this analysis is failing so spectacularly,
>can anyone suggest what's going wrong?
>
>Any instructions or suggestions would be greatly appreciated.
>
>Cheers,
>
>Jon.
>
>_____________________________________________________
>Jonathan Brooks Ph.D. (Research Fellow)
>Magnetic Resonance and Image Analysis Research Centre
>University of Liverpool, Pembroke Place, L69 3BX, UK
>tel: +44 151 794 5629 fax: +44 151 794 5635
>
|