You expect black regions simply because the values are missing in the original
images.
For single subject analyses, it shouldn't be a problem as the realignment,
spatial normalisation etc will (by default - because masking is enabled) put
the same blacked out region in all the data of one subject.
There may be issues for studies that include multiple subjects, as each
subject will have a different pattern of blacked out bits. Really, there
should be a masking procedure that is done between writing spatially
normalised images, and smoothing these data. Most people seem to ignore this
issue though.
Best regards,
-John
On Wednesday 28 December 2005 09:59, Vy (Vee) wrote:
> Dear SPMers,
> I've been trying to preprocess some functional data using SPM2. I noticed
> that after normalizing the data (I coregistered the T1 to the mean image,
> then used the T1 to estimate parameters with respect to a CCHMC template),
> that some superior slices are partially black, as if a black circle or oval
> have blocked out brain data. I also noticed that in some of my data, the
> mean resliced image created during realignment (nothing else was resliced)
> looks like it was "shaved" off at the top of the brain because there was a
> diagonal black line (as if the rest of the brain superiorly was outside the
> field of view). Maybe the normalization process made more superior slices
> black due to interpolation...
>
> So my question is this: How does this affect my analysis? Are those slices
> excluded from my analysis? And how can I remedy this problem?
>
> Just so you know, I used the same bounding box as that of the template.
> I've tried different voxel sizes as well.
>
> Should I be turning off the masking during realignment? Should I be using
> the brainmask that came with the CCHMC template for normalization, or is
> that just for segmentation?
>
> Or can I just continue preprocessing the data?
>
> I'd love to hear your responses. I thank you in advance.
>
> ~V
|