[FieldTrip] ICA question

[Eelke Spaak]

Dear Nenad,

If you do not specify a cfg.channel in your call to
ft_componentanalysis, it will default to 'all'. So, if your data
structure contains EOG channels, they will be used in the
decomposition with no further action required on your part.

If your data is MEG, however, I don't know whether it is wise to
attempt an ICA on the joint data set, including EOG. It is not
standard practice here at the Donders, at least. One possibility to
combine EOG and MEG/ICA data is to correlate your MEG component time
courses with the EOG signal, and then inspect the most highly
correlated components and (if inspection reveals them to be
artifactual) remove them. This hybrid automatic/manual approach seems
to be quite efficient in getting rid of eye artifacts.

But perhaps you can combine them after all; if someone has a more
informed opinion about this please let me know. (Another possibility
is that the story is different for EEG and EOG combined.)

Best,
Eelke

On 29 October 2012 10:04, Nenad Polomac <polomacnenad at gmail.com> wrote:
> Dear all,
>
> I have one question concerning ft_componentanalysis.
>
> I want to calculate ICA for my data in order to detect EOG artifacts and
> remove those ICA components. I have recorded EOG electrodes as well, but I
> am not sure where I should enter channels' label for them. I assume that ICA
> calculation will be more accurate if I provide data from EOG electrodes.
>
> So my question is how to inform ft_componentanalysis about EOG data?
>
> My configuration for the ICA analysis looks like this:
>
>     cfg = [];
>     cfg.method = 'runica';
>     cfg.runica.pca = 90;
>     cfg.runica.maxsteps = 600;
>     cfg.runica.stop = 1e-7;
>     cfg.runica.extended = 1;
>     ica_comp = ft_componentanalysis(cfg, data);
>
> Thank you in advance!
>
> Nenad
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip