[FieldTrip] ft_denoise_synthetic 3rd order grad

Tue Apr 24 10:14:54 CEST 2012

Hi all, A follow-up question on the implementation of 3rd-order gradiometer correction in FT: I am puzzling how one can do both 3-order grad. correction and dft-filtering properly using FT. the logical sequency would be to do ft_denoise_synthetic first, and subsequently dft-filtering using ft_preprocessing. However, in that case it is no longer possible to pad the data when calling ft_preprocessing, because it now processes trial-data (output from ft_denoise_synthetic), and doesn't read in the original dataset anymore. or am I mistaken here? I could see 3 alternatives, which seem to be more like (computationally) suboptimal workarounds: - Apply ft_denoise_synthetic on the entire dataset, then dft-filter the entire dataset, and then run ft_redefinetrial to cut the data into actual trials - Adjust the trial definition such that longer data segments are read in, equivalent to your padding size. Then apply ft_denoise_synthetic on these trials, then dft-filter them, and then run ft_redefinetrial to cut the data into actual trials - Doing the 3-rd order grad. correction first using the DataEditor, and then process it further using FT. Any thoughts? Best, Stan ----- Oorspronkelijk bericht -----
> Van: "Michael Wibral" <michael.wibral at web.de>
> Aan: "Email discussion list for the FieldTrip project"
> <fieldtrip at donders.ru.nl>
> Verzonden: Donderdag 19 april 2012 10:24:57
> Onderwerp: Re: [FieldTrip] ft_denoise_synthetic 3rd order grad
> Hey there,
> sorry for not thinking about it any earlier, indeed I can confirm that
> ft_denoise.. needs the reference channels to function properly. I
> remember having to preprocess in a funny order in old ft versions, to
> do artefcat rejection without the refs and denoising with the refs.
> Personally, I think the code should check whether the refs are there
> and otherwise throw an error. Since ft_denoise.. only concerns CTF
> (and bti??) systems, where the refs have fixed names this could be
> easily implemented, or am I mistaken?
> Best,
> Michael
> Gesendet: Donnerstag, 19. April 2012 um 08:59 Uhr
> Von: "Barbara Haendel" <B.Haendel at gmx.net>
> An: "Email discussion list for the FieldTrip project"
> <fieldtrip at donders.ru.nl>, fieldtrip at donders.ru.nl
> Betreff: Re: [FieldTrip] ft_denoise_synthetic 3rd order grad
> Hey there, thanks a lot for the replies!
> Panagiotis hit the mark. Indeed the reference channels were excluded
> in the first round of preprocessing. May be one could include a short
> note on that in the ft_denoise m-file description since only including
> relevant channels is the normal way to treat data in preprocessing.
> Thanks again,
> Barbara
> -------- Original-Nachricht --------
> > Datum: Mon, 16 Apr 2012 17:53:56 +0200
> > Von: Panagiotis Tsiatsis <Panagiotis.Tsiatsis at Tuebingen.MPG.de>
> > An: Email discussion list for the FieldTrip project
> > <fieldtrip at donders.ru.nl>
> > Betreff: Re: [FieldTrip] ft_denoise_synthetic 3rd order grad
> > Hello girls and guys,
> >
> > It is working for my CTF MEG data - the version that I have is v4692
> > 2011-11-07.
> >
> > The other thing that might cause a problem could be ommiting the
> > reference channels definition before the ft_ preprocessing command
> > that
> > gave you your "dataprepro"
> >
> > i.e maybe you might need something like this
> >
> > cfg = [];
> >
> > ...
> > //various preprocessing parameters
> > ...
> >
> > cfg.channel = {'MEG', 'MEGREF', 'REFGRAD' , 'REFMAG'};
> >
> > dataprepro = ft_preprocessing(cfg);
> >
> >
> > Hope you are lucky :)
> >
> > Best,
> > Panagiotis
> >
> >
> > On 4/16/2012 4:59 PM, Stan van Pelt wrote:
> > > Hi Barbara,
> > >
> > > I can replicate your finding, but am unsure what is causing it.
> > > Maybe
> > > the difference/correction is lower than the Matlab or CTF
> > > precision?
> > > In any case, if the 3rd order gradients were on,
> > > ft_denoise_synthetic
> > > would undo the G3BR-balancing, ergo it toggles.
> > >
> > > Best,
> > > Stan
> > >
> > >
> > > ------------------------------------------------------------------------
> > >
> > > *Van: *"Michael Wibral" <michael.wibral at web.de>
> > > *Aan: *"Email discussion list for the FieldTrip project"
> > > <fieldtrip at donders.ru.nl>
> > > *Verzonden: *Maandag 16 april 2012 16:46:27
> > > *Onderwerp: *Re: [FieldTrip] ft_denoise_synthetic 3rd order grad
> > >
> > > Hi Barbara,
> > >
> > > are you certain that the original data do not have 3rd gradients
> > > enabled? - looking at them once in dataEditor, switching to 3rd
> > > grads and accidentally forgetting to unclick 'save parameters with
> > > dataset' will make them have 3rd grads...
> > >
> > > When I last used ft_denoise_synthetic it worked pretty much as
> > > described (??).
> > >
> > > Michael
> > >
> > > *Gesendet:* Montag, 16. April 2012 um 15:56 Uhr
> > > *Von:* "Barbara Haendel" <B.Haendel at gmx.net>
> > > *An:* fieldtrip at donders.ru.nl
> > > *Betreff:* [FieldTrip] ft_denoise_synthetic 3rd order grad
> > >
> > > Dear fieldtrip users,
> > >
> > > has anyone experience in using ft_denoise_synthetic? If I try to
> > > calculate 3rd order grads in the below specified way the data does
> > > not change with respect to the input data (which has no synthetic
> > > gradiometer applied). Is there some obvious misunderstanding on my
> > > side of how to use this function or is our environment truly
> > > noiseless :)
> > >
> > > Cheers,
> > > Barbara
> > >
> > > cfg=[];
> > > cfg.gradient ='G3BR';
> > > cfg.trials = 'all';
> > > data_3rdorder = ft_denoise_synthetic(cfg, dataprepro);
> > >
> > >
> > > >> data_3rdorder.grad.balance
> > >
> > > ans =
> > >
> > > G1BR: [1x1 struct]
> > > G2BR: [1x1 struct]
> > > G3BR: [1x1 struct]
> > > current: 'G3BR'
> > > previous: {'none'}
> > >
> > > >> sum(dataprepro.trial{8}(30,:)-data_3rdorder.trial{8}(30,:))
> > >
> > > ans = 0
> > >
> > > --
> > > NEU: FreePhone 3-fach-Flat mit kostenlosem Smartphone!
> > > Jetzt informieren:
> > http://mobile.1und1.de/?ac=OM.PW.PW003K20328T7073a
> > > _______________________________________________
> > > fieldtrip mailing list
> > > fieldtrip at donders.ru.nl
> > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > >
> > >
> > > _______________________________________________
> > > fieldtrip mailing list
> > > fieldtrip at donders.ru.nl
> > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > >
> > >
> > >
> > >
> > > --
> > > Stan van Pelt
> > >
> > > Donders Institute for Brain, Cognition and Behaviour, Radboud
> > > University Nijmegen
> > > Kapittelweg 29, 6525 EN Nijmegen, Netherlands
> > > E-mail: stan.vanpelt at donders.ru.nl
> > > Website: www.ru.nl/donders/
> > > Phone: (+31) (0)24 36 10981
> > > Fax: (+31) (0)24 36 10989
> > >
> > >
> > > _______________________________________________
> > > fieldtrip mailing list
> > > fieldtrip at donders.ru.nl
> > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> >
> >
> > --
> > Panagiotis S. Tsiatsis
> > Max Planck Institute for Biological Cybernetics
> > Cognitive NeuroImaging Group
> > Tuebingen, Germany
> >
> --
> NEU: FreePhone 3-fach-Flat mit kostenlosem Smartphone!
> Jetzt informieren: http://mobile.1und1.de/?ac=OM.PW.PW003K20328T7073a
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
-- Stan van Pelt Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Kapittelweg 29, 6525 EN Nijmegen, Netherlands E-mail: stan.vanpelt at donders.ru.nl Website: www.ru.nl/donders/ Phone: (+31) (0)24 36 10981 Fax: (+31) (0)24 36 10989
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20120424/04db32b1/attachment-0002.html>