Dear Luca,
Thanks for your reply.
The native MRI 01_1_mr.nii has dimensions 256x220x256, Vox size 0.977x0.8x0.977, origin 121 145 150.
I perform the segmentation with default settings except from enabling to save the deformation fields.
Then I go to "normalise write" and add the deformation field iy_01_1_mr.nii and as image to write I add for example the MNI template avg152T1.nii. Then as bounding box I add [NaN NaN NaN; NaN NaN NaN] and as a voxel size [NaN NaN NaN], although it is not 100% accurate in this example I use the nearest neighbor interpolation (the actual images that I normalise are atlases).
The result is the wavg152T1.nii, which has dimensions 256x220x256, Vox size -0.977x0.8x0.977, origin 136 145 150.
As you can observe the result is suboptimal. The dimensions are the same but the origin is off and the Vox size is not the same given the negative sign in one dimension.
If I do another reslice with the mri defining the space I get the correct origin and voxel size but is it needed to do a reslice after a normalization?
Am I doing something totally wrong?
Cheers,
Konstantinos
|