Dear Idil,
> I have run into a problem concerning the display of SPM8 results of a single subject (healthy subject). The results are shifted downwards and very small in size compared to the glass brain (see file "mask_sx.bmp" for illustration). I checked the mask image created in the result folder after first level design matrix estimation and it was way smaller than the glass brain and different from other 18 subjects' masks.
> I checked all the preprocessing steps. Coreg, segmentation, normalization all worked and final images are all aligned with the templates provided by SPM.
> One note about this subject might be worth to mention here: the origin of her anatomy was manually adjusted to the AC-PC plane before segmentation else the segmentation created weird results (i.e half of the brain is black, scattered gray scale inflated brain etc)
If I understand correctly, you have checked that:
1) The subject's functional images are properly aligned with each
other and the mean functional image;
2) The mean functional image and the structural image for this subject
are aligned (if you realigned the structural image to match the
template, you would also have needed apply the same adjustments to all
of the functional images);
3) Segmentation seemed to work reasonably;
4) The normalized structural and functional images are in good
alignment with SPM's templates (suggesting normalization worked
properly).
If this is all true, then it could be something a bit off about the
signal for this subject. The mask SPM uses is based on the signal
intensity in a voxel across the timeseries. Generally this mask has
the effect of excluding non-brain regions (that have low signal). So,
there's a chance that there is some increased signal dropout for this
subject (or something odd about the scanner that day). Though off the
top of my head, it seems a bit odd that it appears to equally affect
all parts of the brain—I can't think of a good reason for this. I
would have a look at the mask along with the mean functional and
structural, and see if you can get a sense of what regions are cut
off. You could do this using CheckReg, or perhaps overlaying them in
something like MRIcron.
As has already been suggested, one way to deal with this is to disable
SPM's masking, and instead provide an explicit mask to ignore the
non-brain voxels. However, it's worth considering that if there is
something odd about this subject's data, you may just not want to
include them in the end (if you can't sort out what the problem is).
If nothing else works, I would also suggest re-processing this subject
from the beginning, just to make sure that everything worked as it
should.
Good luck!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|