Different results using multitapers and wavelet

Martin Bleichner bleichner.martin at GMAIL.COM
Mon Dec 1 16:17:40 CET 2008


Hi Ingrid,

thanks for the tip, but it does not solve the problem.
I tried now the method that I am normally using for the low frequencies
(hanning) for my high frequency range and get results comparable to the
wavelet. See below.

Any other suggestions?

best
Martin



cfg.output ='pow'
cfg.method='mtmconvol'
cfg.tapers='hanning'
cfg.toi=[-0.5:0.01:2.5]
cfg.foi=[55:2:95]
cfg.t_ftimwin=4./cfgLow.foi


2008/12/1 Ingrid Nieuwenhuis <ingrid.nieuwenhuis at fcdonders.ru.nl>

>  Hi Martin,
>
>
>
> If you use multitapers (or any frequency analysis) you have to take care
> that you have an integer amount of cycles in your timewindow. Since you use
> a fixed timewindow of 0.2 seconds, your frequency steps should be a multiple
> of 5, so cfg.foi should be 55:5:95 or either you should use a cfg.timwin
> which is dependent on the cfg.foi (fi 7 cycles, just as in your wavelets).
> See the tutorial on the FieldTrip website for more explanation. (fieldtrip »
> documentation » tutorial »Time-frequency Analysis Using Multitapers and
> Wavelets)
>
>
>
> Best Ingrid
>  ------------------------------
>
> *From:* FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL] *On
> Behalf Of *Martin Bleichner
> *Sent:* Monday, December 01, 2008 2:48 PM
> *To:* FIELDTRIP at NIC.SURFNET.NL
> *Subject:* [FIELDTRIP] Different results using multitapers and wavelet
>
>
>
> hi there,
>
> I am using multitapers as well as wavelets to compute my time frequency
> spectrum of ECoG data (50Hz-90Hz). Recently I realized some discrepancy
> using these two methods, until than I found that these to method lead to
> comparable results.
>
> In one of our datasets we found very large differences between the two
> methods. When I looked closer at the problem I found out that the two
> methods lead to similar results when the effect (increase in power) is
> large, while they deviate quite a bit when the effect is small.
> In that case  i get  a kind of 'checkerboard'  like time frequency spectrum
> for the mutlitapers.
> I also compared the output of the frequency analysis with the bandpassed
> raw signal and found that the wavelet transform was the better match. A
>
> re there any prerequisites for multitaper use? Does anybody has some
> experience with this effect? Do I do something wrong?
> Below you can find the cfg settings I am using
>
> Multitaper
> cfg.output ='pow'
> cfg.method='mtmconvol'
> cfg.tapers='dpss'
> cfg.toi=[-0.5:0.01:2.5]
> cfg.foi=[55:2:95]
> cfg.t_ftimwin=ones(size(cfg.foi))*0.2;
> cfg.tapsmofrq=ones(size(cfg.foi))*15;
>
>
> Wavelet
> cfg.method='wltconvol';
> cfg.width=7;
> cfg.foi=[55:2:95]
> cfg.toi=[-0.5:0.01:2.5]
>
> Thanks
>
> Martin
>
> ----------------------------------
>
> The aim of this list is to facilitate the discussion between users of the
> FieldTrip toolbox, to share experiences and to discuss new ideas for MEG and
> EEG analysis.
>
> http://listserv.surfnet.nl/archives/fieldtrip.html
>
> http://www.ru.nl/fcdonders/fieldtrip/
>
> ----------------------------------
>
> The aim of this list is to facilitate the discussion between users of the
> FieldTrip toolbox, to share experiences and to discuss new ideas for MEG and
> EEG analysis.
>
> http://listserv.surfnet.nl/archives/fieldtrip.html
>
> http://www.ru.nl/fcdonders/fieldtrip/
>

----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20081201/d6f76eea/attachment-0001.html>


More information about the fieldtrip mailing list