External Email - Use Caution
Dear friends,
I would like to proof reliability of a source reconstructed with dSPM. How
to apply correction for multiple comparison (FDR or cluster-size) at the
source level? How to setup a threshold for significant activation?
Thank you very much for your help,
Irina
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200204/f7e7c9c2/attachment-0001.html
External Email - Use Caution
hi Irina,
you should give us for details. What conditions / statistical test you
have in mind?
is it at the group level ?
you can find on the mne.tools website some examples eg using
spatio-temporal cluster level stats.
HTH
Alex
External Email - Use Caution
Hi Alex,
Thank you very much for your reply,
I meant a group averaged evoked response (ERP/ERF) to auditory stimulation
(when subjects are performing an active task). The individual dSPMs were
morphed beforehand to a standard template. The group average shows strong
activation within the auditory cortex, and weaker activation in another
region of interest. The second region can be seen at a lower threshold,
that can be adjusted using the interactive scale. So, I would like to find
out whether this second activation is significant. Is there an objective
way to set up the threshold for the significant activation (versus
baseline)? Is it necessary to perform the FDR correction to the group
averaged evoked responses (like we do for fMRI data)?
Thank you very much for your help!
Irina.
External Email - Use Caution
Sorry for posting this question again...
Thank you very much in advance!
External Email - Use Caution
Hi Irina,
I am sure other MNE users have suggestions on this but I can answer with
what
MNE-Python can offer at this stage which is to do: FDR correction on a
paired t-test
or spatio-temporal cluster level stats in source space.
Alex