Head movement correction w/o MaxFilter

💬 Support & Discussions
, ,
1

I have a MEG dataset that was recorded in 10 different short 5 minute sessions. Head position was recorded using 4 tracking electrodes. I do not have the MRI data.

I would like to use sensor-space decoding, and therefore it seems sensible to apply movement correction to the data, such that the head position is approximally equal in all 10 sessions.

However, the data was already published with compensators applied and the whitening matrix (?) is missing do undo those corrections. Therefore, MaxFilter refuses to run on the data.

Is there any other way in Python-MNE to apply movement correction? I found some rather old comments about mne_map_data, but most links were dead and it seems like it has only been implemented in C-MNE.

What would you advise in this case?

2

maybe https://mne.discourse.group/t/how-to-realign-the-ctf-meg-runs-to-a-common-head-position-using-maxwell-filter/4727/12 can help?

A

3

Perfekt, thanks! Somehow I wasn’t finding this thread with my searches…

However, it does seem to introduce some downscaling to my data. (Left the target raw, right, the mapped raw). Also tested it on a decoding approach and decoding accuracy reduces quite significantly (5 sessions, around 20% reduction in accuracy on a visual decoding task when using mapped data and original, non-corrected data).

https://i.ibb.co/rtpK15W/Capture.png

Anyway, if I apply the transform to the target raw (raw_to) as well, everything seems to be in the same scale.