ERD/ERS on source space data

✉️ Mailing List Archive (read-only)
1

External Email - Use Caution

Hi!

Is it possible to compute ERS/ERD maps based on source space instead of
sensor space?

Best,
Karin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20181010/25ba5f09/attachment.html

2

Hi Karin,

Yes, you need to pick few labels in the source space to estimate ERS/ESD
on it.

Please look at the following example:

https://martinos.org/mne/stable/auto_examples/time_frequency/plot_source_label_time_frequency.html

Thanks

Sheraz

3

External Email - Use Caution

You mean basically using source_induced_power instead o tfr_multitaper?

Thanks for answering,
Karin

Den tors 11 okt. 2018 kl 00:55 skrev Sheraz Khan, PhD <
sheraz at nmr.mgh.harvard.edu>:

Hi Karin,

Yes, you need to pick few labels in the source space to estimate ERS/ESD
on it.

Please look at the following example:

Page Redirection

Thanks

Sheraz

> External Email - Use Caution
>
> Hi!
>
> Is it possible to compute ERS/ERD maps based on source space instead of
> sensor space?
>
> Best,
> Karin
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> Mne_analysis Info Page

-------------------------
Sheraz Khan, M.Eng, Ph.D.
Instructor in Radiology

Athinoula A. Martinos Center for Biomedical Imaging
Massachusetts General Hospital
Harvard Medical School

McGovern Institute for Brain Research
Massachusetts Institute of Technology

Tel: +1 617-643-5634
Fax: +1 617-948-5966
Email: sheraz at nmr.mgh.harvard.edu
       sheraz at mit.edu
Web: http://sheraz.mit.edu
_______________________________________________
Mne_analysis mailing list
Mne_analysis at nmr.mgh.harvard.edu
Mne_analysis Info Page

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20181011/fec9aacd/attachment.html

4

Yes, in addition, once you extract the source space data from the label,
you can also use tfr_multitaper on the label time series.

HTH

Sheraz

5

External Email - Use Caution

hi,

i faced a similar question some time ago. there is one thing to keep in
mind when doing (time)frequency transforms on source data:

when data is projected to source space, it is projected to three
components per source: one for each spatial dimension. these three
components are then normally combined by taking the norm. this step
includes squaring the data, so all samples that were negative are now
positive. this is mostly ok when staying in the time-domain (although
one must be aware of it, but it will not mess up anything). however, it
essentially doubles the frequencies of all oscillations in your data, so
it is highly problematic for (time)frequency transforms.

mne python provides a few ways to avoid this problem:

1. use source_induced_power. this function calculates the TFR for all
    three components of every source and combines the power values
    afterwards.
2. restrict the amount of spatial components per source to 1 when
    setting up the source model. i do not know if that is possible in
    mne python.
3. restrict apply_inverse to return only one component.
    pick_ori='normal' would work well for EEG but not for MEG, though
    because it is mostly sensitive to the two tangential components.
4. set pick_ori to 'vector' which should return all three spatial
    components per source, apply whatever algorithm(s) you need and
    combine them later yourself.

i hope, this is all correct. if not, please correct me!

cheers,
thomas