I am not sure this topic really belong to Anatomist section specifically but it seems the closest section to post into.
Here is a description of what I encounter:
I have two niftii volumes produced by the HCP consortium, a T1 data and the associated diffusion data. According to HCP description paper and scripts
diffusion data is registered into the T1 space (T1 itself has been rigidly registered to MNI_152 space in a previous step).
When I load both volumes into Anatomist and load their referential I observe indeed that both volumes have a transformation toward common MNI space.
When I display data in the common referential, both volumes seems to be aligned. However when I look at the mm coordinates into the MNI space of the linked cursor it does not have the same exact coordinates in both images (see captures enclosed). In addition, I noticed that the difference seems to be in coordinate axis corresponding to the displayed axis (see captures enclosed). Since I want to put meshes from T1 into the diffusion referential I need
the coordinates to correspond.
So wondering if it was only a display problem or a registration problems (only one coordinate is wrong because other may be fixed due to selected view) I checked for affines using nibabel and regarding the affines in seems volumes are well aligned. So I verified affine display into anatomist (the aims_mm to mni_mm affine) by recomputing it using numpy and checking that I obtain well the nibabel affine in the end. Calculation seems to be accurate.
Then using the different transformations I checked if going from a voxel into T1 data I would obtain exactly a voxel into diffusion data (the python script is enclosed)
What I see is that I have a small shift on the last two coordinates. I don't know if it is an error of estimation of the aims_mm to ras_mm, a weird bug
or something I misunderstand in coordinates system.
If you have any hint

Alex