John,
Thanks so much. That works very well.
-David
-----Original Message-----
From: John Ashburner [mailto:[log in to unmask]]
Sent: Friday, August 11, 2000 4:04 AM
To: [log in to unmask]; [log in to unmask]
Subject: Re: Making & Saving Percent Change Images
| Another (hopefully not too naive) question from a new user...
|
| I'd like to create, for each individual subject, a subtraction image
that
| reflects %change in normalized rCBF. Thus, instead of t-values, the pixel
| values of this image would be numbers reflecting change above or below
| average whole brain. In my particular case, I have two baselines and two
| activations, so I'd like to create the percent change subtraction image
of:
| (i1+i3)/2 - (i2+i4)/2.
|
| Is there a way to easily accomplish this in SPM? As far as I can tell,
| proportional scaling only comes as part of a process that produces a
| statistical parametric map image, and I don't see anything in the image
| calculator that would enable me to perform this step separately (i.e.,
take
| an image, normalize each pixel by whole brain average, and then do the
| subtractions).
In Matlab, you can obtain the "globals" for each image by:
V = spm_vol(spm_get(4,'*.img'));
gl1 = spm_global(V(1))
gl2 = spm_global(V(2))
gl3 = spm_global(V(3))
gl4 = spm_global(V(4))
Then these can be plugged into the ImCalc expression by:
(i1/gl1+i3/gl3)/2 - (i2/gl2+i4/gl4)/2
I think you actually need to enter the values of the globals rather than
the variable names.
Best regards,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|