[FieldTrip] Cluster-based analysis - different steps in different tutorials

[Diego Lozano-Soldevilla]

Hi Iris,
Thank you for pointing that out. Let's see whether I'm able to help you out:

1.      When does one need to do the grandaverage?
Use ft_timelockgrandaverage function when you want to compute the grand
average across participants and conditions. You decide when you want to
compute this, there's no strict order. In the tutorials, as you saw, this
is done
http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock/#reading-in-preprocessing-timelockanalysis-planar-gradient-and-grandaveraging
and
http://www.fieldtriptoolbox.org/tutorial/eventrelatedstatistics/#reading-in-preprocessed-and-time-locked-data-in-planar-gradient-format-and-grand-averaged-data

2.      If the grandaverage is done after the permutation analysis, what do
we write in the neighbours’ structure? The output from ft_timelockanalysis
(i.e. allsubjFC)?
You can compute the neighbours structure before the ft_*grandaverage (
http://www.fieldtriptoolbox.org/faq/how_does_ft_prepare_neighbours_work/).
In the tutorial data, as all participants were recorded in MEG (same
helmet) and the same coils were consistently recorded in all participants,
you can compute the neighbour structure in one of your subjects or in the
grand average. You can also use templates (
http://www.fieldtriptoolbox.org/template/neighbours/).

Critically, if you're dealing with your own EEG data, you have to make sure
all your subjects have the same amount of electrodes to compute the
neighbour structure before using ft_*statistics.m. In EEG, often you
encounter  participants with nonusable electrodes (they broke/didn't
contact scalp) and you either reject them or interpolate them (
http://www.fieldtriptoolbox.org/workshop/madrid2019/tutorial_cleaning/#interpolating-bad-channels
).

I hope that helps!

Diego


On Fri, 15 Mar 2019 at 12:35, MILÁN MAILLO, IRIS <iris.milan at upf.edu> wrote:

>
> Hi everyone,
>
> I am a newbie in Fieldtrip and I'm trying to do a cluster-based
> permutation test within subjects. I am looking at these two tutorials:
> http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock/
> http://www.fieldtriptoolbox.org/tutorial/eventrelatedstatistics/
>
> In the two of them, it is said that the steps to follow are (in order):
> ft_timelockanalysis, ft_timelockgrandaverage, and ft_timelockstatistics.
>
> However, in
> http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock/
> they seem to perform ft_timelockanalysis followed by ft_timelockstatistics
> and after that, they do ft_timelockgrandaverage (this is not what is
> proposed in the other tutorial). I don't know if this makes a big
> difference or not, but inconsistencies across tutorials are confusing.
> Moreover, in the within-subjects part (
> http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock/),
> it is said that the configuration in neighbours creation is the same as in
> the between-trials:
>
>
>
> “cfg.neighbours = neighbours;  % same as defined for the between-trials
> experiment”
>
> Nevertheless, in the between-trials experiment, they use the GA_FC to do
> the neighbours structure, but in the within-subjects trial it is not
> possible to do so, because the grandaverage is only done after the
> permutation test. So what we need to do?
>
> In sum, my questions are:
>
> 1.      When does one need to do the grandaverage?
>
> 2.      If the grandaverage is done after the permutation analysis, what
> do we write in the neighbours’ structure? The output from
> ft_timelockanalysis (i.e. allsubjFC)?
>
>
>
> Thanks in advance,
>
>
>
> Iris
> --
> *Iris Milán Maillo*
>
> PhD Student
> Speech Production and Bilingualism (SPB) <https://www.upf.edu/web/spb>
> Center for Brain and Cognition (CBC) <http://cbc.upf.edu/cbc.upf.edu/>
>
> Carrer de Ramon Trias Fargas, 25-27, 08005 Barcelona | Edifici Merce
> Rodoreda, (room 24.311)
> _______________________________________________
> 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/20190315/64899094/attachment-0002.html>
-------------- next part --------------
_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202