[FieldTrip] Connectivity comparison across conditions

Jose Rebola jrebola at gmail.com
Mon Jul 13 12:28:12 CEST 2015


Hi Jan (and Vitória),

Thank you for the explanation!

Are the methods you proposed implemented in Fieldtrip in any way?

José

On Fri, Jul 10, 2015 at 5:12 PM, Vitória Piai <v.piai.research at gmail.com>
wrote:

>  Hi Jose,
>
> Just adding two helpful references that further clarify Jan-Mathijs's
> point:
>
> "Fourier-, Hilbert- and wavelet-based signal analysis: are they really
> different approaches?"
> http://www.ncbi.nlm.nih.gov/pubmed/15262077
>
> "Comparison of Hilbert transform and wavelet methods for the analysis of
> neuronal synchrony"
> http://www.sciencedirect.com/science/article/pii/S0165027001003727
>
> Cheers, Vitoria
>
>  On 7/9/2015 7:24 AM, Schoffelen, J.M. (Jan Mathijs) wrote:
>
> Jose,
>
>    Regarding the time-resolved PLV, isn't it different to do it with
> mtmconvol, which in my view provides a TFR, than with a time signal based
> on the Hilbert transform which estimates instantaneous phase.
>
>  My doubt is that the time-resolved PLV will need three or four cycles
> for good spectral estimation, and two things can happen:
> 1 - I analyse non-overlapping chunks and get a low number of time-points.
> 2 - I analyse highly overlapping segments and thus a lot of PLV
> estimations but not really independent from each other and therefore
> biasing the comparison to other conditions
>
>  With the Hilbert derived transform I would theoretically have as many
> independent estimates of the PLV, right?
>
>  Is there a caveat in my line of thought? I might be missing something
> here…
>
>
>  The thing you are missing here, is that the Hilbert transform also
> integrates over time to get an ‘instantaneous’ estimate of phase and
> amplitude (in other words, there’s nothing instantaneous about it). The
> difference with Fourier based methods is that you explicitly have to set
> the time window when doing a short window FFT, and for the Hilbert
> transform (which by the way in MATLAB is computed through a Fourier
> transform as far as I remember) one only can guess about how independent
> the time samples are from one sample to the next. Note, that temporal
> overlap is not so much an issue when doing statistical inference by means
> of permutation tests.
>
>  Best,
> Jan-Mathijs
>
>
>   Jan-Mathijs Schoffelen, MD PhD, Senior researcher
>
> Max Planck Institute for Psycholinguistics
> Donders Centre for Cognitive Neuroimaging
>
> E-mail: j.schoffelen at donders.ru.nl
> Telephone: +31-24-3614793
>
> http://www.hettaligebrein.nl
> http://www.fieldtriptoolbox.org
>
>
>
> _______________________________________________
> fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
>
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150713/6324dc59/attachment.html>


More information about the fieldtrip mailing list