Dear Johan,

I'm not sure whether using individual T1 without individual electrode
position measurement provides any advantages over the template. My
guess would be that it's unlikely but the answer probably also depends
on the exact analysis pipeline at least to some extent. Your
coregistration looks OK. If some of the electrodes are a bit inside
the head, it's not a problem as their locations are projected to the
head surface at the forward modelling step which you can verify if you
display the forward model. I think the blue circles correspond to the
fiducials. You are somehow using all the electrodes as fiducials but 3
should be enough. In any case, I don't think it's a problem as long as
the forward model display looks OK.

Best,

Vladimir

On Thu, Jan 18, 2024 at 5:55 AM Johan <[log in to unmask]> wrote:
>
> Dear SPM users,
>
> I am working with EEG data and would like to use SPM to get the signal from different anatomical sources using the forward/inverse modeling procedure, and the OSL pipeline scripts later on. A critical step in this procedure is the co-registration of the electrodes to the subject's T1 image.
>
> While there is a Template-based approach, I believe that using the subject's own T1 would improve source localization. Is this correct?
>
> I did some work to do this co-registration without any fiducial information, but just re-calculating standardized electrode positions based on the transformation from MNI T1 -> Subject T1 space. The following figure is the result of that work. I'd like an assurance (or a rejection) of the co-registration result I've come up with?
>
>
>
>
> Does the co-registration using the subject's own T1 figure (right figure) look right to you? On the left is the co-registration using the MNI Template brain. Am I right into assuming the diamonds are the initial positions, and the circles are the final positions? Or should I think about it differently?
>
> Many thanks for your insights
> Johan
>
>
>
>
> --
>
> Some more information:
>
> I do not have any fiducials or anything; just the electrode positions, from the electrode position file from the EEG cap manufacturer (ANT-Neuro).
>
> Assuming the electrodes are somewhat closely aligned to MNI space, I thought I could figure out how to transform the electrode positions (that are well aligned to the MNI template image) to the subject's own T1 scan. So, I used rotation, scaling and offset information from co-registering the MNI T1 to the subject's own T1, to re-calculate calculate the position(s), and then to try match electrodes to the MRI, by specifying each electrode as a point, and using 'use headshape points' in the matlabbatch GUI. I compare this approach with the 'template' approach and compare:
>
> Some electrodes are within the skin, but seem to be pushed 'out' of the red lineout. I think that the standard MNI template head has a different head size to brain size ratio, than my particular test subject.
>
> If this approach seems to work, and assuming that the cap is put on correctly to the subject, this might make it possible to do co-registration of EEG to the head just based on the T1 without measurement of the Fiducials. Fiducials were sometimes measured, but in many cases were not reliable.
>
>
>
>
>