[FieldTrip] ICA decomposition & backprojection

[jan-mathijs schoffelen]

Hi Patrick,

> 1.)    automatically modifies the leadfield matrix?

No, not automatically. However, if you construct the leadfields, and use the grad-structure from the ICA-cleaned data, then you should be fine, because the subspace projection is applied both to the time courses and to the gradiometer description. 

> 2.)    the unmixing (and mixing?) matrix is saved in the new data structure somewhere (to evtl make use of it at later analysis steps)?

No, the unmixing and mixing matrices are stored in the comp-structure, which you obtain after ft_componentanalysis. They are not passed on to data structures obtained from processing steps downstream. If you want to keep track of this, you should save the comp-structure, or at least the mixing and unmixing matrices (along with the topolabel and label fields). The component time courses are to some extent redundant, because these can be easily computed from the data.

> 3.)    automatically stores the information how many and what ICA components were subtracted from the data in my new data structure.

Yes, this info is stored in data.cfg.component, where data is the output structure after calling ft_rejectcomponent.

> Would you generally discourage to apply ICA in the preprocessing when planning source reconstruction and connectivity analysis?

No, I wouldn't. Yet it should be applied with care, and I would indeed check the results of the analysis pipeline with and without ICA-cleaning.

Cheers,
JM


Jan-Mathijs Schoffelen, MD PhD 

Donders Institute for Brain, Cognition and Behaviour, 
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands

Max Planck Institute for Psycholinguistics,
Nijmegen, The Netherlands

J.Schoffelen at donders.ru.nl
Telephone: +31-24-3614793

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20121121/2dbb9154/attachment-0002.html>