Sensory/anatomy registration for two different recordings of the same subject

Discussions
,
1

Hello, I have two MEG recording sessions for the same subject. These sessions have been recorded one right after the other.
After selecting "review raw file" for each of them, I need to perform the sensory/anatomy registration separately. Is there a way to do it only once for the two files?
This is especially important if the registration is manually adjusted.
Thank you very much.

2

Even in this situation, it is possible that subject's head is in a slightly different position inside the MEG helmet. This has as consequence that the same MEG sensor may not record brain activity from exactly the same brain region in each session. Check more details here:
https://neuroimage.usc.edu/brainstorm/Tutorials/ChannelFile#Multiple_runs_and_head_positions

If the head movements between sessions are going to be ignored, it is possible to use same registration for both sessions by copying the channel file from one session to the other:

  1. Import Session1 and perform the sensor registration
  2. Import Session2, leave the default registration (it is going to be overwritten)
  3. Copy the channel file from Session1 to Session2 (can be done with Ctrl-C and Ctrl-V shortcuts)
3

Thank you very much for your help!
One fundamental question, please: I know that MEGIN (MEG) systems have a "motion correction" feature. The way I understand it, once we do "motion correction", we don't need to align the MEG sensors. I thought the above process is only important for aligning fiducials/scalp points with MRI data. Isn't this the case?
(This is why I thought that doing this registration twice for the same subject was redundant.)
Thank you!

4

More or less, considering that you have digitized head points (scalp and anatomical fiducials), these two things happen in the process:

  1. Alignment of the head points and the anatomy (fiducials in MRI and head surface).

  2. Alignment of the head points to MEG sensors. This is performed using the digitized head localization coils, and their location with respect to the dewar (which is acquired at the start of the recordings).

For step 1, as you mentioned, it is not needed to re-do it, as the anatomy and the digitized head points are the same. This is the reason you can use the .pos file for both sessions.

For step 2, however, the head position (and the attached head localization coils) with respect to the MEG may have changed due to head movement. This information is saved in the channel file. Thus if the channel file is copied, the initial position of the head in the MEG is also copied.

When you adjust manually, only the step 1 is adjusted.

Currently, @Marc.Lalancette is working in a better way so the step 1 could be shared across the same subject in back to back recordings.

5

Hello,

(For anybody else reading this, this explanation assumes digitized head points and fiducials are present.)

First, regarding the "motion correction" of MEGIN, it doesn't really change anything to what Brainstorm does. That procedure is more about correcting the changes of position during the recording, but Brainstorm still needs the (now fixed) position of the head vs the helmet/sensors which is saved in each recording. So from the perspective of Brainstorm, it's the same as usual where it uses only the initial position.

But you are correct that manually editing the coregistration with the 3d figure ends up changing the relative position of the digitized head points (scalp, anatomical fiducials, and head coils) with respect to the anatomy (head surface extracted from MRI), and thus this should be the same for all recordings within the same session. The "link" between 1) the digitized head coils and 2) the head coil positions measured by the MEG system and saved in each recording (possibly modified by MEGIN's motion correction), is "automatic/implied" and cannot be modified.

As Raymundo mentioned, I've been dealing with this same issue with our OMEGA data, where we have to manually edit the head position because the digitized head points are often not good (I won't get into that issue here). What I've been doing is editing one recording manually, then using that to update the MRI anatomical fiducials to match the digitized ones. This then automatically applies the correction to other recordings. But one has to be careful not to have other transformations (like refine using head points) applied as well. Eventually this process will be available, but it may take some time still.

In the meantime, another perhaps safer and easier to implement option would be to have a script or process to copy the manual adjustment transformation from one recording to the others of the same session. In the channel file, there's a field for MEG transformations, and one with labels, and same for EEG transformations. From memory, I think for MEG it's something like:

  1. native to dewar
  2. dewar to Brainstorm/SCS
  3. refine with head points (if it was done)
  4. manual adjustment (if it was done)

So the manual adjustment transformation would always be last and it's therefore not too difficult to copy it and add it to the other channel files (both the transformation and the label, and both to MEG and EEG fields if present). You'd have to make sure the "refine" transform is present in all of them as well, or just copy and overwrite it in others too.

Best,
Marc