| I have a question concerning the segmentation option
| on SPM'96 for WindowsNT.
| I like to segment a single MRI with a resolution of
| 256*256 and 53 slices. Do I need a special template or
| probability image for this case ?
You shouldn't need any special templates, unless the image
you are trying to segment has a much different contrast to
the existing template images.
| If I want to normalize or realign the image with T1
| then occurs always an error. When I use the segment
| option the images for white, gray, CSF and others are
| always empty. Is it necessary to realign the image
| before I can start segmentation ? What is the correct
| sequenz for the segmentation operation ?
What happens if you use the <Display> button with the image?
Do you see anything or is it completely garbled? If you
do see an image, are the voxel sizes correct, and is the
origin somewhere near to the AC. Also, does the display
option show the transverse slice at the lower left, with
the front of the brain shown at the top?
In order to segment an image, SPM needs to know some form
of mapping between the image and the prior probability images.
If the images are spatially normalised, then this is fine,
otherwise, SPM estimates the best affine mapping between the
image and the template(s) you specify. By knowing this spatial
mapping, the segmentation routine is able to overlay the priors
on to the image in order to assist the segmentation.
|
| I have also problems with the header. I converted the
| image von dicom to analyze. SPM#96 NT can't read that.
| I always have to fix the header. Do SPM#96 NT require
| a special version of the analyze format ?
There is one extra field that SPM uses, and that is the ORIGIN
field. See spm_hread.m to get an idea where it is.
Good luck,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|