MRI-MEG Co-registration and refining head positions

Discussions
1

Dear community,
I am working on MEG data (channel file .fif format).
I have a couple of questions to clear about the topic.

  1. I have a subject anatomy created by using freesurfer. When I import the anatomy, I have to assign fiducial points, manually. But I worry about the misalignment of those points.
  2. Our MEG recording also has 7 digitized head points but I could not refine the head points since the number of head points are very low. I am looking for refining the head positions in order to compensate for different head positions for multiple runs. Any solutions and/or recommendations???

Thank you.

2

  1. The points you have to select when you import the anatomy are the anatomical fiducials that you clicked with the polhemus pen before you started recording the MEG. These are anatomical points, they are not related with the position of the head in the MEG and should not be modified to “fix” this position.

  2. Indeed, with 7 head points, you should not try to use the menu “Refined with head points”. Unless the registration looks really off, I recommend you leave the registration the way it is based only on the NAS/LPA/RPA points.
    The position of the head between runs might be different between runs, this is normal, do not try to change the position the sensors to make it similar across the runs, this would not fix the data… If the head of the subject a lot moved a lot between runs, you have 2 options: use MaxFilter (Elekta only) to register the runs before importing in Brainstorm, or average across runs only in source space.
    http://neuroimage.usc.edu/brainstorm/Tutorials/ChannelFile#Multiple_runs_and_head_positions

3

Dear Francois,

Thank you so much for your response.

As you mentioned, I considered the anatomical fiducial points which can be found in Elekta channel file (.fif).

I checked those points on the following steps:

  1. Import raw MEG data to the brainstorm.
  2. Export channel file (say filename is ChanFile) to the Matlab.
  3. Open ChanFile.SCS.

Then it shows the following co-ordinates for the fiducial positions:
NAS: [0.0929,0,0]
LPA: [-0.0014,0.0753,0]
RPA: [0.0014, -0.0753,0]

My problem here is that I could not transform these co-ordinates into equivalent MRI co-ordinates and use those points as a precise fiducials.

I am sorry, if my question is ambiguous.

4

You need to define INDEPENDENTLY the position of the NAS/LPA/RPA in the anatomy (MRI coordinate system) and in the MEG (subject coordinate system), so that the program can register the two together.
You cannot convert the position of these points in the MEG coordinate system to the MRI before computing the transformation matrix between the two coordinate systems.

Please refer to the corresponding tutorial:
http://neuroimage.usc.edu/brainstorm/Tutorials/ChannelFile

5

Hi Francois,
Is there any update for "
Method 3: Co-register properly the runs using the process Standardize > Co-register MEG runs: "
https://neuroimage.usc.edu/brainstorm/Tutorials/ChannelFile#Multiple_runs_and_head_positions

6

Hi Rasmita,
I suggest you ask your question in more detail in a new post. What are you hoping to do here? The option of transforming the data to artificially "align" the MEG sensors between runs would only make sense if you're interested in analyzing MEG at the sensor level only, which is not very common. What "update" are you asking about?
Cheers,
Marc

7

Hi Marc,
Yes - I have described it here: Co-registration of multiple runs
Thanks,
Rasmita