Remove muscle artifacts using ICA

jan-mathijs schoffelen jan.schoffelen at DONDERS.RU.NL
Fri Oct 29 09:52:57 CEST 2010


Dear Marc,

Your figures seem to be missing, so it is hard to judge what the
artifacts look like exactly. Could it be that one of your head
localization coils was switched on througout the measurement?
In general, if your goal is to do source localization, I would not try
to fix ugly channels, but just omit them from the sensor-array,
because there will be plenty of sensors left.
The fixing operation (whatever way it is done, e.g. nearest neighbour
interpolation, ICA etc) involves replacing each channel's estimate by
a linear combination of a subset of/all other channels. You have to
keep in mind that the solution to the forward model (i.e. the
leadfields for the sources you want to estimate) have to take the same
linear operation into account in order to give correct results.
As such, irrespective of the fact that the noisy channels are on the
edge of the array, interpolation does not really make sense, because
you are not really improving the quality of your total signal array.
Also, in this case, I don't expect that rejecting the independent
component capturing the artifact will be that beneficial, because most
likely the spatial topography of this component of this component will
be confined to the three bad guys, with more or less random loadings
on the rest of the channels. Did you check whether the artifact is
present at the level of the reference sensors? If that's the case, you
could consider applying the cfw and afw (compute fixed weights, and
apply fixed weights) utilities from the 4D software.

Best wishes

Jan-Mathijs


On Oct 28, 2010, at 7:11 PM, Marc Recasens wrote:

> Dear all,
>
> I have quite a naive question.
> I'm processing some MEG (4-D) datasets in order to use source
> location methods afterwards. One of my concerns is that I have some
> channels (3 in a row) with a steady high frequency artifact >50Hz (I
> thought it is muscle activity, However it is very tonic and present
> during the whole recording) which is within my frequencies of
> interest. This can be seen in the attached figures: timelocked
> responses bandpass filtered between 15 and 150 Hz, and time-
> frequency activity between 50 and 100 Hz.
> As the artefactual channels are put altogether in the right edge of
> the sensor array (A148, A147 and A146) interpolation may not be a
> suitable method to eliminate those artefactual channels. (?)
>
> I was wondering whether it is possible to correct those artifacts
> using ICA in such a way similar to ECG artifact removal using
> component analysis, that is, by identifying  and remove those
> components in the source analysis that explain the high-frequency
> artifacts present in some of my channels.
>
> Thanks a lot.
>
>
>
>
>
>
>
>
>
>
> --
> Marc Recasens
> Tel.: +34 639 24 15 98
> --------------------------------------------------------------------------- You
>  are receiving this message because you are subscribed to
> the FieldTrip list. The aim of this list is to facilitate the
> discussion
> between users of the FieldTrip toolbox, to share experiences
> and to discuss new ideas for MEG and EEG analysis.
> See also http://listserv.surfnet.nl/archives/fieldtrip.html
> and http://www.ru.nl/neuroimaging/fieldtrip.
> ---------------------------------------------------------------------------
>

Dr. J.M. (Jan-Mathijs) Schoffelen
Donders Institute for Brain, Cognition and Behaviour,
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands
J.Schoffelen at donders.ru.nl
Telephone: 0031-24-3614793


---------------------------------------------------------------------------
You are receiving this message because you are subscribed to
the  FieldTrip list. The aim of this list is to facilitate the discussion
between  users of the FieldTrip  toolbox, to share experiences
and to discuss  new ideas for MEG and EEG analysis.
See also http://listserv.surfnet.nl/archives/fieldtrip.html
and http://www.ru.nl/neuroimaging/fieldtrip.
---------------------------------------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20101029/cf74fad2/attachment-0001.html>


More information about the fieldtrip mailing list