[FieldTrip] automatic IC rejection

Joseph Dien jdien07 at mac.com
Sat Dec 1 00:08:26 CET 2018


I have a comprehensive automated artifact correction routine (MAAC) implemented in my EP Toolkit, which you may find helpful.  It follows the principle that different artifacts have different characteristics and so different algorithms will be most effective for each one.  I’m in the middle of writing it up.

https://sourceforge.net/projects/erppcatoolkit/

Joe

> On Nov 30, 2018, at 12:08, Jason Taylor <jason.taylor at manchester.ac.uk> wrote:
> 
> Hi Aitor,
> 
> No, not that I know of. I generally use a hacky combination of SPM, fieldtrip, and EEGLAB functions, but if you've already run ICA, you could accomplish what I suggested with some standard matlab functions. 
> 
> Best wishes,
> Jason
> 
> -----Original Message-----
> From: Aitor Egurtzegi [mailto:aitor.martinezegurcegui at uzh.ch] 
> Sent: 30 November 2018 13:43
> To: Jason Taylor; FieldTrip discussion list
> Subject: Re: [FieldTrip] automatic IC rejection
> 
> Dear Jason,
> 
> Thanks a lot for your reply. Is there a Fieldtrip method already 
> implemented to run such temporal correlation? or would I have to do the 
> implementation in raw Matlab?
> 
> Thanks in advance,
> Aitor
> 
> On 11/30/18 2:05 AM, Jason Taylor wrote:
>> Hi Aitor,
>> 
>> If you have an 'objective' measure of the artefact you're trying to remove (e.g., VEOG for blinks), a relatively straightforward method is to run a temporal correlation between each IC's activation time-course and the artefact channel's time-course. You can then reject any IC with a correlation higher than some threshold, or with a Z-score (r value relative to the distribution of IC r values) above some threshold. This tends to work very well for identifying blinks, and fairly well for eye-movements (*EOG), and can work for pulse artefact if you have recorded ECG. To avoid spurious correlations due to high-frequency noise, you can filter (e.g., 1 to 30 Hz) the component and artefact signals before correlating them (but obviously go back to the original unfiltered signals to continue with your analysis).
>> 
>> Best wishes,
>> Jason
>> 
>> 
>> -----Original Message-----
>> From: fieldtrip [mailto:fieldtrip-bounces at science.ru.nl] On Behalf Of David Schubring
>> Sent: 28 November 2018 12:47
>> To: fieldtrip at science.ru.nl; aitor.martinezegurcegui at uzh.ch
>> Subject: Re: [FieldTrip] automatic IC rejection
>> 
>> Dear Aitor,
>> 
>> the closest thing I know of for a data-driven approach of selecting
>> independent components is COMPASS, quote:
>> 
>> "COMPASS is a MATLAB and EEGLAB based algorithm with the purpose of
>> providing the user with a convenient technique for automatic Independent
>> Component (IC) selection with respect to the contributions of the ICs to
>> a certain ERP."
>> 
>> Link to the toolbox:
>> 
>> http://53450283.de.strato-hosting.eu/jrw/lab/e_compass.htm
>> 
>> Paper:
>> 
>> Wessel, J. R., & Ullsperger, M. (2011). Selection of independent
>> components representing event-related brain potentials: a data-driven
>> approach for greater objectivity. Neuroimage, 54(3), 2105-2115.
>> https://doi.org/10.1016/j.neuroimage.2010.10.033
>> 
>> I have only theoretical experience with the toolbox as I only learned
>> about it in a workshop and did not yet have the time to test and
>> implement it in my personal FieldTrip workflow (even though it is on my
>> ever growing to-do list). So far it looked like a useful thing to try
>> out to me, especially as code can better be reproduced than "personal
>> judgement".
>> 
>> Best,
>> David
>> 
>> Am 28.11.2018 um 10:49 schrieb Aitor Egurtzegi:
>>> Dear researchers at Fieldtrip,
>>> 
>>> 
>>> In order to make my work more reproducible, I would like to
>>> automatically reject ICs instead of doing visual inspection and
>>> rejection of the components. Unfortunately, I haven't found any
>>> documentation for such thing. Is there a way to do it in Fieldtrip?
>>> 
>>> Best,
>>> Aitor
>>> 
>>> _______________________________________________
>>> 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

--------------------------------------------------------------------------------

Joseph Dien, PhD
Senior Research Scientist
Department of Human Development and Quantitative Methodology
University of Maryland, College Park
http://joedien.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20181130/437f48a1/attachment.html>


More information about the fieldtrip mailing list