> We have acquired 6 slices of ASL images and wish to normalize.  We
> coregister these and a structural scan to a whole brain EPI acquired at
> the same sitting. 2 questions,
> 1) What is the better way to proceed, i) normalize the coregistered
> structural to the T1 template and apply _sn.mat to the coregisterd
> functional scans or  ii) normalize the EPI and apply _sn.mat to
> coregistered functional scans, or is there no difference?

There is likely to be a difference - but the better option may be difficult to 
predict.  One of the main factors relates to how much distortion you have in 
your EPI data.  If it is very distorted, then a good registration with the 
structural scan may not be possible - in which case any spatial normalisation 
parameters derived from the structural won't be especially accurate when 
applied to the EPI.  Other factors are the contrast and artifacts in your 
data.  Because spatial normalisation relies on minimising the mean squared 
difference between an image and template, then the contrast in the data 
should be similar for optimal results.


> 2) if / when normalizing the coregistered structural do we select the r*
> image or  the original structural,  - in light of previous posting [Re:
> normalization of few slices,Mon, 4 Dec 2000 15:26:50 +0000 (you don't
> need to use the r* image created by coregistration, as the
> coregistration transformation is added to the structural MRI .mat file)]?

Select the original image (where the relative orientation is encoded by the 
.mat file). If you try the Check Reg button with the original data after 
coregistration, then you should see that these are also in register.  This is 
done via the .mat files, rather  than by reslicing the data themselves.

The reason for chosing the original image is that resliced images sometimes 
contain regions where there is no data in the original image.  These are 
assigned a value of zero, which the spatial normalisation would try to 
distort because there would be no corresponding zero regions in the template 
image.

Best regards,
-John