Hello,
> After I analyzed fMRI data of one subject, I found the activation exceeds a
> lot out of the glass brain. I went back to check the images after each
> preprocessing step, and found after normalization, top of the brain have
> been clipped. I attached the JPEG image of Check Reg result: the two images
> are one of the normalized volume and the EPI.nii template. My set up of
> normalization (estimate and write) is: the source image is the individual's
> structural image which has been aligned to AC-PC. Images to write are the
> realigned and coregistered fMRI scans (during coregistration, fMRI scans are
> registered to
> individual's structural image), and the template is T1.nii.
Is there any chance that the functional images got misaligned from the
structural image somewhere along the way? I.e., if you use CheckReg,
can you verify that the mean functional image, some randomly-chosen
functional image, and the structural image for a subject all line up?
I haven't run into normalization issues like you describe before,
except where I've made a mistake in the coregistration between image
types.
Does this happen for multiple subjects?
> P.S. Is it advisable to normalize fMRI images to EPI.nii template, without
> using individual's structural image?
Some people have reported better results with this; as far as I know,
most people get quite good results using the structural image. You
might also want to try using the Segment option to estimate the
normalization parameters, which gave better normalization results than
the standard SPM normalization in the recent Klein et al. paper
(http://dx.doi.org/10.1016/j.neuroimage.2008.12.037).
Best regards,
Jonathan
|