I am trying to get a T1w image aligned with both the SPM MNI space and the fMRI image that is aligned to the MNI space.
I use SPM to co-register the structural and functional images together, and use spm for spatial normalization to the TPM defined space. Doing this results in a 256x256x256 fMRI image.
I then take the warp and apply it to my structural image (done with SPM). This produces a w prepended image that should be warped to the same space as the fMRI image. However, this warped T1w image has a size of 157x189x156. I understand that this warped image has a much tighter bounding box to reduce data size. However, how can I expand this bounding box, and importantly, how do I expand this bounding box and make sure that this warped t1 lines up 1:1 with the warped fMRI volume (where the voxel indices in the two images would perfectly overlap).
If the two images are aligned with each other, then all you would need to do is regrid one image to the other. You could use something like mrgrid (mrgrid — MRtrix 3.0 documentation). But I would be careful in that it sounds like you are either going to lower the resolution of your T1 image by a lot (losing data) or upsample your fmri by a lot (leading to unnecessary data interpolation). Based on your fMRI image resolution, it appears you might have already upsampled your BOLD image unless you were able to acquire it with sub-mm resolution.
This is taken from Justin O’brien’s Youtube channel, but this is what the post warping t1 looks like. It is the same resolution (1mm^3) as the warped fMRI, so I think this is a padding issue. Expanding the box to 256x256x256 would likely fix this issue, I think, but I don’t know how to pad it correctly. The warped fMRI has far more empty margins around the brain, it’s not so tightly bound.
