I have used the MRI registration to edit all of my channel files (128 EEG electrodes + 3 fiducial points) into appropriate positions. For the most part, they were already exactly where they should be, and seemed sensible, so I “projected electrodes on to scalp surface” prior to generating my BEM surfaces. Everything looks good to my eye…did the button that projects the electrodes to the scalp essentially warp the head shape to the default anatomy? Or should I still go through the process of “Digitized head points > Warp > Deform default anatomy to fit these points”
As a side note, After I finished using the MRI registration on all my channel files, I was forced to go into the actual file, and select “Delete” under Type for the fiducial points, as an error messages kept popping up that matrices did not match up when attempting to calculate noise covariance. I still have the original channel files with the fiducial points, but I’m hoping this is just a technical issue that won’t present problems with further analysis? As far as I can tell, the fiducial points have served their purpose up to this point and won’t be of further use anyways…thoughts?
To summarize, you have two head shapes from different subjects: the head surface from the Colin27 template and the electrodes locations digitized on your subject’s head.
To match the two in an EEG study, you have two options to make them match:
either you transform the electrodes to fit the template head surface (EEG-MRI registration > Project electrodes on scalp surface)
or you transform the MRI and surfaces to fit the head shape defined by the electrodes positions (Digitized head points > Warp > Deform to fit)
It’s one or the other. If the case of #2, you should not edit the positions of the electrodes before, except for solid transformations (rotations+translations of the whole set of electrodes) to align roughly the subject’s head shape with the Colin27 brain.
I’m not sure what gives the best results in EEG. The approach #1 is probably easier. The warping of default anatomy (option #2) was developed more for MEG studies. In MEG, the positions of the sensors are fixed, so we cannot do something like “project sensors on the scalp surface”.
If you want to go for this option #2 anyway, you can get better results by adding more head points than just the electrodes. You can do this with the Brainstorm digitizer interface: http://neuroimage.usc.edu/brainstorm/Tutorials/TutDigitize
In the database, the channel file must have exactly the same number of entries in the .Channel field as the recordings (number of rows of the .F field in the data files).
The fiducials are not supposed to be imported as “electrodes” in the channel file, they should be saved in the field .SCS in the channel file and they should not appear as points that you can edit using the menu “MRI registration > Edit”.
If they are saved as electrodes, you should delete them from the channel file immediately after importing it, either with the menu “MRI registration > Edit” or with the menu “Edit channel file”.
Please let me know if you have further questions.
Francois
Dear Francois,
I have one difficulty related to same topic.
You have mentioned above “The fiducials are not supposed to be imported as “electrodes” in the channel file, they should be saved in the field .SCS in the channel file”. If my case they do not appear as points that you can edit using the menu “MRI registration > Edit” but they appear in " Edit channel file" as Misc type. How I can ensure that they are stored as .SCS or whether Misc type corresponds to .SCS.
Further my EEG data (64 channel) and channel file (64+3 fiducials+ 1 REF) are in different file. When channel file is imported its show 68 and EEG file is 64 so its shows mismatch of my number of channels while opening EEG data. Solution is to delete 3 fiducials + 1 ref points and it works but I think its inaccurate as BST take into account that remaining 4 are fiducials and error should not appear. May be you can guide it anything wrong in the process.
What if order of electrodes label (1 to 64) if different in EEG data file and channel file ? Is it matched automatically based on label of channel or same order should be maintained in both files ?
Thanks in advance for reply.
Regards
Eshwar Ghumare
What is the file format of the file you are trying to import? What software generated it?
Could you post it on this forum or send it to me by email?
It is usually easier to edit the electrodes file manually before importing it into Brainstorm rather than trying to understand how to modify it afterwards.
You can re-write the text file in a very simple format, so that it is read correctly and without any ambiguity in Brainstorm (X, Y, Z coordinates in centimeters):
NameElec1 X Y Z
NameElec2 X Y Z
NameElec3 X Y Z
...
Nasion X Y Z
LPA X Y Z
RPA X Y Z
Then you import this file using the file format “EEG: ASCII: Name, XYZ (.)”
You need to re-order the electrodes in this file to match the order in your recordings. There is nothing that matches the name of the electrodes in the EEG recordings and the channel file, you are responsible for importing an electrode file that labels correctly the channels in the recordings. Skip the references and everything else that is not explicitly saved in your recordings.
The naming of the fiducials is correct, you don’t need to rename them. FidNz / FidT9 / FidT10 are recognized correctly as being the nasion/left ear/right ear. You can see that if you right-click on the imported channel file > File > View file contents, at the bottom of the file the structure SCS is filled correctly
You probably have to remove a few entries from this list (eg. Ref, HEOG*, VEOG*) to match the list of channels in your recordings
You may need to re-order the electrodes in this file to match your recordings
You can import this .sfp file with the file format “EEG: ASCII: Name,XYZ” or “EEG: BESA (*.sfp;…)”