PLV in label in source space

Dear all,

When calculating the phase-locking value (PLV) from n trials in a label in
source space, I'm wondering if it is more relevant to:
- spatially averaging the trials across the label, then calculated the PLV.
- calculating the PLV on each source then averaging the PLVs across the
label.

I compared the two methods, and things change a lot (the PLV is larger when
the average is done at the level of the trials than when it's done at the
level of the PLVs). I would have the feeling that averaging the trials
first is more relevant, especially if using svd, so that the averaged
signal is more robust. Could someone confirm/contradict my feeling?

Thanks in advance,
Laetitia G.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20140919/0c7ef6b5/attachment.html

Dear Laetitia,

Generally I find that spatially averaging across the label, then calculating the PLV works better.

Best wishes,
Avniel

hi,

thanks for sharing these insights. Any suggestion to make this straightforward
and update/deprecate source_induced_power is welcome.

Laetitia if you can/want to share a gist it would be great.

Alex

Dear Alex and Laetitia,

One thing to mention is that my experience is that the statistical
significance does not change much either way. This has to do with the fact
that averaging first boosts the SNR of the signal going into the PLV
calculation, increasing the PLV values. If one calculates the PLV for each
source, the individual PLV values are lower (lower SNR for each estimate),
but because you have averaged many together, your confidence in the
resulting PLV is increased. You can do this comparison using monte carlo
or permutation testing, the p-values should not change much as long as you
perform all of the same procedures on the permuted data as you did for the
original data.

Best wishes,
Avniel

Well, sure, I can share my code, but I'm using the deprecated function
induced_power... I didn't want to use the new tfr_morlet function because
it remains in sensor space.
Here is my script:
https://gist.github.com/anonymous/361c7ca1db3355c73618

Thanks Avniel for your precision! But do you think that it remains true
when you're using svd averaging? Moreover, I'm interested in an index
calculated as followed: [PLV(condition 1) -
PLV(condition1+2)]* [PLV(condition 2) - PLV(condition1+2)]. Thus this index
is proportional to PLV?, and maybe it's why, when I compared the two index
pattern, I saw big differences (I didn't try the stat yet).

Thanks again!
Laetitia

2014-09-19 17:58 GMT+02:00 Ghuman, Avniel <ghumana at upmc.edu>:

Dear Alex and Laetitia,

One thing to mention is that my experience is that the statistical
significance does not change much either way. This has to do with the fact
that averaging first boosts the SNR of the signal going into the PLV
calculation, increasing the PLV values. If one calculates the PLV for each
source, the individual PLV values are lower (lower SNR for each estimate),
but because you have averaged many together, your confidence in the
resulting PLV is increased. You can do this comparison using monte carlo
or permutation testing, the p-values should not change much as long as you
perform all of the same procedures on the permuted data as you did for the
original data.

Best wishes,
Avniel

>hi,
>
>thanks for sharing these insights. Any suggestion to make this
>straightforward
>and update/deprecate source_induced_power is welcome.
>
>Laetitia if you can/want to share a gist it would be great.
>
>Alex
>
>
>> Dear Laetitia,
>>
>> Generally I find that spatially averaging across the label, then
>>calculating the PLV works better.
>>
>> Best wishes,
>> Avniel
>>
>> From: Laetitia Grabot
>><laetitia.grabot at gmail.com<mailto:laetitia.grabot at gmail.com>>
>> Reply-To: Discussion and support forum for the users of MNE Software
>><mne_analysis at nmr.mgh.harvard.edu<mailto:
mne_analysis at nmr.mgh.harvard.edu
>>>>
>> Date: Friday, September 19, 2014 10:27 AM
>> To:
>>"mne_analysis at nmr.mgh.harvard.edu<mailto:
mne_analysis at nmr.mgh.harvard.edu
>>>"
>>><mne_analysis at nmr.mgh.harvard.edu<mailto:
mne_analysis at nmr.mgh.harvard.ed
>>>>>
>> Subject: [Mne_analysis] PLV in label in source space
>>
>> Dear all,
>>
>> When calculating the phase-locking value (PLV) from n trials in a label
>>in source space, I'm wondering if it is more relevant to:
>> - spatially averaging the trials across the label, then calculated the
>>PLV.
>> - calculating the PLV on each source then averaging the PLVs across the
>>label.
>>
>> I compared the two methods, and things change a lot (the PLV is larger
>>when the average is done at the level of the trials than when it's done
>>at the level of the PLVs). I would have the feeling that averaging the
>>trials first is more relevant, especially if using svd, so that the
>>averaged signal is more robust. Could someone confirm/contradict my
>>feeling?
>>
>> Thanks in advance,
>> Laetitia G.
>>
>>
>>
>>
>> _______________________________________________
>> Mne_analysis mailing list
>> Mne_analysis at nmr.mgh.harvard.edu
>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>
>>
>> The information in this e-mail is intended only for the person to whom
>>it is
>> addressed. If you believe this e-mail was sent to you in error and the
>>e-mail
>> contains patient information, please contact the Partners Compliance
>>HelpLine at
>> http://www.partners.org/complianceline . If the e-mail was sent to you
>>in error
>> but does not contain patient information, please contact the sender and
>>properly
>> dispose of the e-mail.
>>
>
>_______________________________________________
>Mne_analysis mailing list
>Mne_analysis at nmr.mgh.harvard.edu
>https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis

_______________________________________________
Mne_analysis mailing list
Mne_analysis at nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20140919/1ee63594/attachment.html