help
help > RE: Different dimensions after normalization SPM
Jun 8, 2015 04:06 PM | Alfonso Nieto-Castanon - Boston University
RE: Different dimensions after normalization SPM
Hi Laura,
As far as I can tell the reason why you are getting different dimensions/resolutions in your normalized volumes is because you are explicitly entering a different target resolution for each subject. I would recommend to simply leave the default [2 2 2] voxel size when resampling during normalization and that should result in (as you expect) MNI-space volumes that have the same dimensions/resolution across all subjects.
Also, if you prefer to use CONN preprocessing pipeline, your procedure (again, if I am understanding correctly) is referred to as "indirect normalization" (in "indirect normalization" you first coregister your functional to your anatomical volumes, then you normalize the anatomical volumes and apply the same estimated normalizing-transformation to the functionals), so if you prefer you may simply select that pipeline and CONN will perform the realignment/coregistration/normalization procedure automatically.
Hope this helps
Alfonso
Originally posted by Laura E:
As far as I can tell the reason why you are getting different dimensions/resolutions in your normalized volumes is because you are explicitly entering a different target resolution for each subject. I would recommend to simply leave the default [2 2 2] voxel size when resampling during normalization and that should result in (as you expect) MNI-space volumes that have the same dimensions/resolution across all subjects.
Also, if you prefer to use CONN preprocessing pipeline, your procedure (again, if I am understanding correctly) is referred to as "indirect normalization" (in "indirect normalization" you first coregister your functional to your anatomical volumes, then you normalize the anatomical volumes and apply the same estimated normalizing-transformation to the functionals), so if you prefer you may simply select that pipeline and CONN will perform the realignment/coregistration/normalization procedure automatically.
Hope this helps
Alfonso
Originally posted by Laura E:
Hi everyone,
I have a question with respect to normalization in SPM. I thought that when we normalise to MNI space, the nii resultant had the same dimension for all studies of patients independently of dimensions of the original image.
For example,
I have a T1 of subject 1 with dimensions 256x256x186 and [0.94 0.94 1.00 ] size of voxels , we realign the image with the images SWAN, first we choose T1 and second SWAN and we choose the option "Register to first" and the resultant images with prefix r, T1 and SWAN have the dimensions 256x256x186 and [0.94 0.94 1.00 ]. After we normalise with option "estimate and write" we choose in "image to align" r_t1.nii image and we select r_swan.nii in "Images to write". We change the "voxels size" to [0.94 0.94 1.00 ], the rest of values by default. The resultant image wr_swan.nii have a dimensions 167x201x156 with size of voxels [0.94 0.94 1.00 ].
And now the T1 of the subject 2 have the following dimensiones 512x512x248 and [0.47 0.47 0.5] size of voxels and following the same procedure that with the subject 1, the wr_swan.nii resultant has a dimensions of 333x402x156 and [0.47 0.47 1.00] size of voxels.
Is this correct and normal? I need to compare the images SWAN of the differents patients.
I would have to put the same size of voxels for all patients? For example the default values [2 2 2] ?
Is normal that when we normalise to MNI space we have images with different dimensions if the original images has different dimensions or size of voxels?
I'm using PET & VBM in SPM 12.
Thanks to everyone in advance
Best regards
Laura
I have a question with respect to normalization in SPM. I thought that when we normalise to MNI space, the nii resultant had the same dimension for all studies of patients independently of dimensions of the original image.
For example,
I have a T1 of subject 1 with dimensions 256x256x186 and [0.94 0.94 1.00 ] size of voxels , we realign the image with the images SWAN, first we choose T1 and second SWAN and we choose the option "Register to first" and the resultant images with prefix r, T1 and SWAN have the dimensions 256x256x186 and [0.94 0.94 1.00 ]. After we normalise with option "estimate and write" we choose in "image to align" r_t1.nii image and we select r_swan.nii in "Images to write". We change the "voxels size" to [0.94 0.94 1.00 ], the rest of values by default. The resultant image wr_swan.nii have a dimensions 167x201x156 with size of voxels [0.94 0.94 1.00 ].
And now the T1 of the subject 2 have the following dimensiones 512x512x248 and [0.47 0.47 0.5] size of voxels and following the same procedure that with the subject 1, the wr_swan.nii resultant has a dimensions of 333x402x156 and [0.47 0.47 1.00] size of voxels.
Is this correct and normal? I need to compare the images SWAN of the differents patients.
I would have to put the same size of voxels for all patients? For example the default values [2 2 2] ?
Is normal that when we normalise to MNI space we have images with different dimensions if the original images has different dimensions or size of voxels?
I'm using PET & VBM in SPM 12.
Thanks to everyone in advance
Best regards
Laura
Threaded View
| Title | Author | Date |
|---|---|---|
| Laura E | Jun 8, 2015 | |
| Alfonso Nieto-Castanon | Jun 8, 2015 | |
| Laura E | Jun 15, 2015 | |