I'm analysing a dataset in MNE-Python that has been acquired in a CTF scanner. I also have access to individual MR's for each subject. Both one in the native coordinate system (Niftii) and one converted to the CTF space. Since the data comes from an publicly accessible database, the MR's have been defaced, and therefore it is not possible to correctly get the nasion fiducial correctly placed.
My attempt at solving this has been to run the FreeSurfer pipeline on the MR in the CTF space. Then to get it in the Neuromag Space I've rotated it 90 degrees around the z-axis and translated it 40 cm upwards. This looks reasonable, but now I've gotten worried that this won't work since the FreeSurfer origin is based on the ACPC convention and not between the auricular points as in Neuromag. Does this mean that my co-registration below is likely to be wrong, or is it right to be in the Neuromag coordinate system?
Here's my transformation matrix to go from CTF (left below) to Neuromag (right below)
[cid:part1.F7596179.D5AB9AE9 at ki.se]
[cid:part2.26D2E596.8AE582B7 at ki.se]
Left: before rotation and translation. Right: after rotation and translation.
{
"DigitizedHeadPoints": "sub-A2002_headshape.pos",
"DigitizedHeadPointsCoordinateSystem": "CTF",
"DigitizedHeadPointsCoordinateSystemDescription": "CTF coordinates relative to the nasion and left and right pre-auricular points",
"DigitizedHeadPointsCoordinateUnits": "cm",
"FiducialsDescription": "Coregistration was initially done using the approximate position of the head localizer coils in the anatomical MRI and subsequently refined using the full headshape measured with the Polhemus. The positions of the head localizer coils at the ear canals (relative to the Polhemus coordinates) is included in the headshape file. CTF coordinates in the MEG data and in the coregistered anatomical MRI are specified relative to the localizer coils at nasion and left and right ear canal.",
"HeadCoilCoordinateSystem": "CTF",
"HeadCoilCoordinateSystemDescription": "CTF coordinates relative to the localizer coils, with the localizer coils at nasion and left and right ear canal",
"HeadCoilCoordinateUnits": "cm",
"HeadCoilCoordinates": {
"nasion": [10.5092,0,0],
"left_ear": [-0.014438,7.03839,0],
"right_ear": [0.014438,-7.03839,0]
},
"IntendedFor": "anat/sub-A2002_space-CTF_T1w.nii",
"MEGCoordinateSystem": "CTF",
"MEGCoordinateSystemDescription": "CTF coordinates relative to the localizer coils, with the localizer coils at nasion and left and right ear canal",
"MEGCoordinateUnits": "cm?
}
So the anatomical landmarks (nasion, left and right ear canal) are specified and their voxel indices can be looked up in the corresponding anatomical MRI, even though the nose is not visible any more. This should allow you to redo the full coregistration procedure using MNE Python. Note that the _headshape.pos file is also shared.
cheers
Robert