Hi Claire
> 1. Thank you so much for your detailed response, Jonathan! Following up the response, would it be problematic if I use 'segment' to bias correct the structural image (yes it will produce seg.sn.mat file but I don't plan to use this) and then use 'normalize (estimate&write) to extract a parameter file (using the bias-corrected T1 as the source image and functional images as the ones to write)?? I am asking whether it is OK because during realignment for functional images I plan to choose only a mean image to be resliced and thus T1 and functional may not be aligned with one another...
First, to answer your actual question :) , it should be perfectly fine
to normalize the bias-corrected version as you suggest.
However, can I ask why you don't want to use the *seg_sn.mat file to
normalize? In general this seems to be more accurate; for example,
there is at least one quantitative comparison I know of
(http://dx.doi.org/10.1016/j.neuroimage.2008.12.037) that suggests the
unified segmentation approach is more accurate.
In any case, regardless, you will need your T1 to be aligned to all
the functional images, because whatever normalization parameters you
obtain for your structural image (either through 'normalise: estimate'
or through segmentation) will then be applied to all the functional
images; this won't work if they're not lined up to begin with. :)
So, you want to do the following (and see, for example, Chapter 28:
Auditory fMRI data, in the SPM manual, for more details):
1. Realign (create mean image only) (now all functional images are
lined up with each other; even though they are not resliced the
updated positions are saved in the image header)
2. Coregister (structural to mean functional) (now all functional
images are lined up with your T1) (note that generally you want the
mean EPI to be the reference image, which remains stationary, and the
structural to be the source image, which will be moved to match)
3. Segment
4. Normalise: Write (using *seg_sn.mat file from segmentation)
5. Smooth
> 2. Another question (not related to the above): after coregistration, I am supposed to have a 'mutual information coregistration' image appeared on the graphics window automatically but it doesn't seem to appear...what could be wrong?
Sorry, I don't have any ideas on this one. It would help those with
more knowledge than me to know what version of SPM you are running
(and whether you have the latest updates), what version of Matlab, and
on what type of computer. Do you have problems with any other SPM
graphics?
Hope this helps,
Jonathan
|