The spatial normalisation in SPM is for estimating the parameters that
warp an image to match a template of the same modality. Once estimated,
then the same warp can be applied to any other images that are coregistered
with it. After warping, the image should be in a different position to how
it was originally - hopefully matching the template.
In order to estimate warps from a structural image and apply them to
functional images, then they must first be in register. You can do
this with the <Coregister> button. If the default coregistration method
does not work for you, then you can use the mutual information version
that is available (use the <Defaults> button to switch). Coregistration
produces (or modifies) a .mat file for the "object" image. The contents
of this .mat file tell SPM how to map between an image and any other
image that has been coregistered with it. You can check what is in register
with what by using the <Check Reg> button.
Once coregistered, spatial normalisation can be used to estimate the warp
for one of the images, and the warp can be applied to any image that is in
register with it (as it takes the .mat files into consideration).
Good luck,
-John
| There is something basic I do not understand about normalization. We have 6
| subjects, 45 epi images and a structural image for each subject; we would like
| to normalize the structural image to the T1 template image and use the
| normalized image to superimpose the activations obtained by statistical
| analysis. This is the procedure I follow:
|
| - Normalize option: the structural image is the image to determine parameters
| and T1.img is the template image. I obtain the struct_sn3d.mat file.
|
| - Defaults option: I choose the default bounding box. As for the voxel size, I
| choose 1 1 3 which is the voxel size for the high resolution structural images.
|
| - Normalize option: I use struct_sn3d.mat as parameter file and write struct.img
| as the image to write normalized.
|
| In this way everything goes through but if I compare the normalized images to
| the original ones I find that, by instance, the nomalized slice for z=0
| corresponds to the not-normalized slice for z=12.
|
| Can anybody help me with this problem?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|