[FieldTrip] ?= redefining data se

Stephen Whitmarsh stephen.whitmarsh at gmail.com
Thu Jan 17 12:25:03 CET 2019 

Hi Martin,

Great. That is what is expected.

The time-units of the original data was in samples. The first 3 columns
defined the trial start, end, and offset, with respect to trial onset, i.e.
stimulation, *defined in samples*.
After segmentation, time is expressed not in samples but with *time
relative to trial onset*, i.e. your .time field.
Start and end samples (i.. reference to original dataset in samples) are
maintained, not in the trialinfo, but in the sampleinfo field of your data.
However, after you e.g. concatenate datasets, resample your data, that
information becomes erroneous, and will be removed by functions such as
append-data.

So as an answer - no, it should be a problem later one.

HTH, good FieldTripping!

Stephen




On Thu, 17 Jan 2019 at 11:43, Martin Rosenfelder <
martin.rosenfelder at uni-ulm.de> wrote:

> Hi Stephen,
>
> I now managed to add the event column to my .trl field. This is now a
> (120x4) matrix. When calling ft_preprocessing, it automatically creates the
> .trialinfo field. Surprisingly, this .trialinfo matrix is now only a
> (1x120) matrix instead of a (4x120) matrix and it contains solely the event
> values. The start/end/offset samples are missing. At the moment I don't
> have any disadvantages from that, but I don't know if this will turn out to
> be problematic.
>
> Best,
> Martin
>
> --
> M.Sc.-Psych. Martin Rosenfelder
> Wissenschaftlicher Mitarbeiter
> Klinische und Biologische Psychologie
> Universität Ulm
> Raum 47.2.259
> +49 731-50 26592
> martin.rosenfelder at uni-ulm.de
>
> Am Mittwoch, 16. Januar 2019 17:35 CET, Stephen Whitmarsh <
> stephen.whitmarsh at gmail.com> schrieb:
>
> > Hi Martin,
> >
> > If I understand correctly: If you add columns* to your .trl field* (so
> > after the first 3 that define start/end/offset samples) which you feed
> into
> > ft_preprocessing, ft_preprocessing will create the .trialinfo field for
> > you, which should track through your analyses.
> >
> > Cheers,
> > Stephen
> >
> > On Wed, 16 Jan 2019 at 17:15, Martin Rosenfelder <
> > martin.rosenfelder at uni-ulm.de> wrote:
> >
> > > Hi Stephen,
> > >
> > > Exactly, preferably, the events should be single values in a struct
> array
> > > or a cell array.
> > > This can then be added to the existing trial array. So far so good.
> > >
> > > However, my dataset misses a .trialinfo field after calling
> > > ft_preprocessing, to which I could add the cell array with the
> condition
> > > labels.
> > > I then added the array with the condition labels to 'mydataset.trial'.
> > > Unfortunately, this seems to confuse the ft_rejectvisual I call
> afterwards
> > > for artifact rejection.
> > > Do I have to create the .trialinfo-field by hand to my dataset?
> > >
> > > Best,
> > > Martin
> > >
> > >
> > > --
> > > M.Sc.-Psych. Martin Rosenfelder
> > > Wissenschaftlicher Mitarbeiter
> > > Klinische und Biologische Psychologie
> > > Universität Ulm
> > > Raum 47.2.259
> > > +49 731-50 26592
> > > martin.rosenfelder at uni-ulm.de
> > >
> > > Am Dienstag, 15. Januar 2019 17:40 CET, Stephen Whitmarsh <
> > > stephen.whitmarsh at gmail.com> schrieb:
> > >
> > > > Hi Martin,
> > > >
> > > > No indeed, your conditions/RT/events should indeed be (re)coded as
> single
> > > > values. This also makes selecting trials etc. much easier with
> logical
> > > > expressions, e.g. you can then simply do cfg.trials =
> > > (data.trialinfo(:,1)
> > > > == 3 && data.trialinfo(:,2) > 0.5, where the first column e.g. is
> your
> > > > condition, and the second RT, to just give an example.
> > > >
> > > > Cheers,
> > > > Stephen
> > > >
> > > > On Tue, 15 Jan 2019 at 17:35, Martin Rosenfelder <
> > > > martin.rosenfelder at uni-ulm.de> wrote:
> > > >
> > > > > Dear Stephen,
> > > > >
> > > > > Thank you very much for the hint on ft_selectdata.
> > > > >
> > > > > I tried to build the structure for .trialinfo as you described it
> in
> > > your
> > > > > reply. I did this creating a structure array with the eventvalues
> as
> > > > > elements.
> > > > > Then I concatenated the .trl matrix and the event value matrix.
> This
> > > > > however failed, since .trl is a double array and the event value
> > > matrix is
> > > > > a struct array.
> > > > > I double-checked that the nr. of rows in the two matrices match
> each
> > > other
> > > > > (120 elements).
> > > > >
> > > > > Is there a way to add the event value matrix (120x1 struct) to the
> .trl
> > > > > matrix (120x3 double)?
> > > > >
> > > > > Best,
> > > > > Martin
> > > > >
> > > > >
> > > > > --
> > > > > M.Sc.-Psych. Martin Rosenfelder
> > > > > Wissenschaftlicher Mitarbeiter
> > > > > Klinische und Biologische Psychologie
> > > > > Universität Ulm
> > > > > Raum 47.2.259
> > > > > +49 731-50 26592
> > > > > martin.rosenfelder at uni-ulm.de
> > > > >
> > > > > Am Montag, 14. Januar 2019 19:01 CET, Stephen Whitmarsh <
> > > > > stephen.whitmarsh at gmail.com> schrieb:
> > > > >
> > > > > > Dear Martin,
> > > > > >
> > > > > > Use ft_selectdata instead of ft_redefinetrial.
> > > > > >
> > > > > >
> > > > >
> > >
> "cfg.previous.previous.previous{1,1}.previous.previous.trialdef.eventvalue(1,2)"
> > > > > > is scary! And I can imagine very inconvenient, as it will move
> > > deeper and
> > > > > > deeper in to infinite previousnessness.
> > > > > > Instead, one of the best 'easter eggs' (i.e. not so
> well-documented
> > > > > > functionality) of FieldTrip is to create extra columns of info in
> > > your
> > > > > .trl
> > > > > > when using preprocessing to epoch your data. These extra columns
> will
> > > > > then
> > > > > > enter into a field .trialinfo of your data. Most if not all
> > > functions,
> > > > > such
> > > > > > as ft_selectdata (selecting trials) will update that field. So I
> > > advice
> > > > > you
> > > > > > to put your eventvalue (as well as RT, response, etc. etc.) as
> > > columns in
> > > > > > your trialinfo (so same nr. of rows as your nr. of trials). In
> this
> > > way,
> > > > > > they will travel with your data, and stay in the same structure
> and
> > > on
> > > > > the
> > > > > > same level whatever happens.
> > > > > >
> > > > > > Cheers,
> > > > > > Stephen
> > > > > >
> > > > > >
> > > > > > On Mon, 14 Jan 2019 at 18:47, Martin Rosenfelder <
> > > > > > martin.rosenfelder at uni-ulm.de> wrote:
> > > > > >
> > > > > > > Dear Fieldtrip community,
> > > > > > >
> > > > > > > I have preprocessed a single dataset with two different
> conditions
> > > > > > > ('Swim', 'Rest'). The conditions are stored in the 'eventvalue'
> > > field
> > > > > of
> > > > > > > the cfg as 1x2 cell array. The event values are stored in the
> > > > > > >
> > > > >
> > >
> 'data_ref.cfg.previous.previous.previous{1,1}.previous.previous.trialdef.eventvalue'
> > > > > > > field of the data set.
> > > > > > > Having done the preprocessing I now would like to do
> > > > > ft_timelockanalysis
> > > > > > > and ft_freqanalysis on the data. Afterwards I statistically
> compare
> > > > > the two
> > > > > > > conditions using ft_timelockstatistics and ft_freqstatistics.
> > > > > > > How can I split the data set according to the event values
> ('Swim',
> > > > > > > 'Rest')? I need the event values to split the data set into
> the two
> > > > > trial
> > > > > > > classes in the ft_timelockanalysis / ft_freqanalysis and to
> compare
> > > > > these
> > > > > > > two conditions.
> > > > > > >
> > > > > > > I tried ft_redefinetrial, but do not know how to define the
> > > cfg.trials
> > > > > > > field of this function.
> > > > > > >
> > > > > > >          % trial redefinition
> > > > > > >
> > > > > > >         % containing only trials in the 'swim' condition
> > > > > > >          cfg.trials = (1,
> > > > > > >
> > > > >
> > >
> data_ref.cfg.previous.previous.previous{1,1}.previous.previous.trialdef.eventvalue(1,1));
> > > > > > >          swim = ft_redefinetrial(cfg,data_ref);
> > > > > > >
> > > > > > >         % containing only trials in the 'resting' condition
> > > > > > >          cfg.trials = (1,
> > > > > > >
> > > > >
> > >
> data_ref.cfg.previous.previous.previous{1,1}.previous.previous.trialdef.eventvalue(1,2));
> > > > > > >          rest = ft_redefinetrial(cfg,data_ref);
> > > > > > >
> > > > > > > I hope the description was quite clear. In case, I can provide
> some
> > > > > more
> > > > > > > lines of code to clarify the issue.
> > > > > > >
> > > > > > > Thank you very much in advance for your advice!
> > > > > > >
> > > > > > > Best regards,
> > > > > > > Martin
> > > > > > >
> > > > > > > --
> > > > > > > M.Sc.-Psych. Martin Rosenfelder
> > > > > > > Wissenschaftlicher Mitarbeiter
> > > > > > > Klinische und Biologische Psychologie
> > > > > > > Universität Ulm
> > > > > > > Raum 47.2.259
> > > > > > > +49 731-50 26592
> > > > > > > martin.rosenfelder at uni-ulm.de
> > > > > > >
> > > > > > >
> > > > > > > _______________________________________________
> > > > > > > fieldtrip mailing list
> > > > > > > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > > > > > > https://doi.org/10.1371/journal.pcbi.1002202
> > > > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > _______________________________________________
> > > > > fieldtrip mailing list
> > > > > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > > > > https://doi.org/10.1371/journal.pcbi.1002202
> > > > >
> > >
> > >
> > >
> > >
> > > _______________________________________________
> > > fieldtrip mailing list
> > > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > > https://doi.org/10.1371/journal.pcbi.1002202
> > >
>
>
>
>
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20190117/af58225a/attachment-0002.html>

More information about the fieldtrip mailing list