Using -trans in maxfilter affects coregistration

[cgohil8]

If you use the -trans option during maxfiltering the placement of the polhemus headshape points/fiducials changes in the coregistration. For example:

Coregistering a file called HC08_rest_tsss.fif (maxfiltered without -trans), we get:

The same subject and structural, however, using the HC08_rest_tsss_trans.fif file, we get:

The head is much closer to the MEG sensors.

It appears the meg to head transformation is different in the fif file if you use -trans in maxfiltering, this can be verified with

import mne

# Maxfiltered without -trans
info = mne.io.read_info("HC08_rest_tsss.fif")
trans = mne.transforms._get_trans(info["dev_head_t"], "meg", "head")
print(trans)

# Maxfiltered with -trans
info = mne.io.read_info("HC08_rest_tsss_trans.fif")
trans = mne.transforms._get_trans(info["dev_head_t"], "meg", "head")
print(trans)

[matsvanes]

Is this still an issue or can we accept that we need to choose to either coregister trans or tsss (and ideally include some documentation on this)

[cgohil8]

Still an issue. Probably not resolvable in the code. The advice is use the _tsss.fif file. This is what's included in the oxford example scripts: https://github.com/OHBA-analysis/osl/tree/examples/examples/oxford.

[matsvanes]

Add warning in the coreg log that the head position might look off whenever trans is used, but this does not affect the source data