Advice on morphing

✉️ Mailing List Archive (read-only)
1

Hello,

I want to use a group-level fMRI map (in fsaverage space) as an fMRI prior
for MNE source localization. I also want to define labels from the fMRI map
that can be used for all subjects.

Previously, I created source spaces for all subjects in each one's
individual anatomical space. I could not find any tools that allow me to
morph these source spaces to fsaverage before doing forward/inverse
solution - it seems most morphing procedures are designed to be applied
after the inverse solution. So it seems my options are:

1. Use the individual source spaces for the forward solution. Morph the
fMRI prior to each subject's individual space to do the inverse solutions.
Define labels on group fMRI map and then morph labels to each individual
space for analysis (or tediously draw labels on each morphed fMRI prior
map).

or

2. Redo the source spaces: morph all subjects to fsaverage space with
mne_setup_source_space and perform forward/inverse solution and all
remaining analysis in fsaverage space.

Any advice or opinions on the best way to proceed? Option 2 seems best as
there is less morphing overall, but I would appreciate input/advice.

Thank you,

Natalie
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20151008/9d0e7c2f/attachment.html

2

Hi Natalie,

IMO
a modified version of option 1 is the best option. Morph the labels to
the individual's space.

Then, instead of morphing the later inverse solutions back to
fsaverage, simply extract the data only from the group defined labels
and run stats on the label data only. This strategy reduces the amount
of morphing and fits your hypothesis that the fMRI labels contain the
activity of interest (which the strong fMRI prior will make happen
anyway). The subsequent morphing would not accomplish anything (except
possibly introduce erroneous smoothing), because the labels are
derived from the average space anyway, so each label is equivalent in
the different participants.

hth
d

3

Hi all,

Note: This is still Natalie, replying from a friend's account because I
didn't get your response to my inbox yet (digest settings), but I wanted to
ask a follow up question.

I want to confirm regarding the prior: when I have the group-level fMRI
prior in fsaverage space as a .w file, I should morph that to each
individual subject's space for use in mne_do_inverse_operator, correct?
(Using mri_surf2surf for instance...)

I will then define labels on the fsaverage fMRI map and morph the labels to
each individual to when I want to extract source estimates for the labels.

Thank you,

Natalie

4

Yes, I think that is the best way to do it in the c-code