The y_* files are actually 5D:  dimx * dimy * dimz * 1 * 3.  It is done this way because the NIfTI specification says that the 4th dimension of a multi-dimensional volume should be time.

Visualising anything other than the first volume using SPM is a bit tricky, but it is possible.  If you try CheckReg, you can elect the first volume and then hit the edit button to allow you to give filenames of the other two volumes.  eg.

/blah/blah/y_blah.nii,1,1
/blah/blah/y_blah.nii,1,2
/blah/blah/y_blah.nii,1,3

The Deformations utility should pretty much do what you want.  You'd use the pullback to bring an image from time 1 into the average space, which should be relatively straightforward.

Best regards,
-John





On 28 January 2015 at 11:36, Marina Papoutsi <[log in to unmask]> wrote:
Dear colleagues,

I recently run pairwise longitudinal registration for two T1w images.
This generated an average image, the jacobian and divergence rate images and the deformation fields (y_*).

According to the spm manual the y_* volume should be a 4D image, containing three volumes one for each x,y,z coordinates. However, in my case, there is only one volume and it looks like a gray-white gradient.
Is this right?

I've tried this with two different types of datasets and the results are similar.
If this is correct, what information does this image contain?

I am interested in the deformations, because I then want to bring some other images (e.g. MT) to the average space to allow me to compare them directly (MT at time 1 vs time 2).

Colleagues suggested that I use the deformations tool.
If I understand correctly, the pullback function brings the warped image (in the case of longitudinal registration I assume this is the averaged image), back to space of the original image, e.g. either at time 1 or time 2.

In contrast, the pushforward warps an image, e.g. T1w at time 1, to the space of the averaged image.
If this is correct, then I can use the pushforward function to bring e.g. the MT at time 1 to the average space.
Can I use the y_* images as they are or do I need to use their inverse?

Thank you in advance for the help,
Marina