Dear Brainstormers,
I would like to perform a grand-average on my MEG dataset (several subjects, several conditions) but I cannot find exact instructions on how to do that.
Following the subjects’ co-registration page
http://neuroimage.usc.edu/brainstorm/Tutorials/CoregisterSubjects
I have projected single-subject anatomies for one condition to the default anatomy, which correctly ended up into the Group analysis folder.
How should I proceed after that to have a single average time course and a single anatomy?
And is it possible to avoid projecting the sources for each condition and rather have a single projected anatomy for each subject?
Thanks for your feedback, or for pointing me to the page where this is explained.
Best,
Marco
Hi Marco,
As you probably know, averaging MEG recordings across subjects is not advised. Because of the different head sizes and positions, one sensor captures the activity of different brain regions, the sensor values cannot be compared directly between subjects.
For averaging across subjects, the recommended procedure is what you’ve been doing: estimate the sources for each subject separately, calculate your subject averages, then project the average source maps on the same template anatomy. When you use the menu “Project sources”, the projected source maps should land in the folder “Group analysis”.
You can re-organize all the files you get in “Group analysis” (create folders, cut-paste or drag-and-drop the files). Then you just have to calculate your grand-averages.
When you want to co-register multiple runs of one subject for calculating a subject average at the sensor level, you can try using the process “Co-register MEG runs”, but this hasn’t been tested thoroughly. It is not adapted to register different subjects together anyways. It is only designed to account for small movements between two runs of the same subject.
What is the step that is not clear for you?
Cheers,
Francois
Thanks Francois it’s all clear now.
Best,
Marco