[FieldTrip] Comparing ICA methods between toolboxes (BESA vs. Fieldtrip)

Wed Feb 3 11:21:21 CET 2016

Hi Andrew,

We discussed this dataset, and we believe that the component which you selected (ICA #22) is most likely a muscle artifact. The ICA components are sorted according to their variance, so the component #22 is one with very little variance. The question is whether you actually need to correct for this artifact. You may find it on only one or few electrodes – may be worth checking. If that is the case, you might consider setting the electrode to bad.

The extended Infomax ICA method is not optimized for correcting muscle artifacts. It lends itself more to the correction of eye movements or cardiac artifacts. In the ICA decomposition that you sent, component #1 looks like an eye movement. 

Components #17 and #19 look like they have mains frequency on them. You can get rid of this by using the notch filter.

Hope this helps.

Best regards,
Harald

From: c.andrew123 at gmail.com [mailto:c.andrew123 at gmail.com] On Behalf Of Andrew Chang
Sent: 29 January 2016 04:26
To: Harald.Bornfleth at besa.de
Cc: FieldTrip discussion list <fieldtrip at science.ru.nl>; Ille Nicole <NIlle at besa.de>
Subject: Re: [FieldTrip] Comparing ICA methods between toolboxes (BESA vs. Fieldtrip)

Hi Harald,

I did the ICA based on the continuos recorded block (~7 minutes), including 100 trials, with PCA option on. The attached file is the ICA topography and waveform (ICA 22). The spectrum showed that the peak power was at ~11Hz.

I suspected it is a neck movement artifact, because the generator seems to be in the front of the neck. Also, I got another similar ICA component from another subject, and the topography seems to be centered at the back of the neck, would it be considered as neck movement too.

Thank you very much once again!

Best,

Andrew

On Mon, Jan 18, 2016 at 7:49 AM, Harald Bornfleth <Harald.Bornfleth at besa.de <mailto:Harald.Bornfleth at besa.de> > wrote:

Dear Andrew,

A couple of questions popped up. 

1)      The neck movement artifact, is this something with low frequency, or is it a muscle artifact?

2)      Did you calculate the ICA on one data screen (e.g. 10 seconds or 20 seconds), or over the whole data set?

If you could send us screen shots showing the ICA wave forms and the accompanying topographies, then we could comment on the classification. (In BESA, you right-click on the label of a component and select “Map topography” to show the topography map of that component.)

I include my colleague and ICA expert Nicole Ille on CC.

Best regards,

Harald

From: c.andrew123 at gmail.com <mailto:c.andrew123 at gmail.com>  [mailto:c.andrew123 at gmail.com <mailto:c.andrew123 at gmail.com> ] On Behalf Of Andrew Chang
Sent: 17 January 2016 19:57
To: Harald.Bornfleth at besa.de <mailto:Harald.Bornfleth at besa.de> ; FieldTrip discussion list <fieldtrip at science.ru.nl <mailto:fieldtrip at science.ru.nl> >
Subject: Re: [FieldTrip] Comparing ICA methods between toolboxes (BESA vs. Fieldtrip)

Hi Dr. Bornfleth,

Thanks for your reply very much. 

I would like to use ICA to remove eye blinks and eye movements, neck movements, as well as cardiac and powerline artifacts. The segment I have is a continuous EEG recording for 10 minutes, which is one block of my experiment including many trials.

Before ICA, the segment have been high-pass (0.3 Hz) and low-pass (100 Hz). The components showing activities around the eyes, the back of the neck, and 60 Hz were removed. As for the cardiac artifact, I it is based on visual inspection on the waveform to see whether it looks like an EKG. 

Any further comments are appreciated!

Best,

Andrew

On Fri, Jan 15, 2016 at 8:35 AM, Harald Bornfleth <Harald.Bornfleth at besa.de <mailto:Harald.Bornfleth at besa.de> > wrote:

Dear Andrew,

Thanks for bringing this up. I can try to shed some light on the ICA method used in BESA Research; however, I can not comment on the implementation in FieldTrip. 

The method behind ICA analysis in BESA Research is the extended infomax ICA algorithm (Lee TW et al.: Independent component analysis using an extended infomax algorithm for mixed sub-Gaussian and super-Gaussian sources. Neural Computation 11(2), 1999, 409-433). This algorithm is very well suited to remove cardiac and powerline artifacts.  Depending on the data, it may also be used to remove ocular (blink) and electrode artifacts. To assess whether your artifact removal is trustworthy more information about your processing steps are needed. What types of artifacts are you trying to correct? What segments are you using to determine the artifact topographies and what segments are you finally correcting? 

Best regards,

Harald

Dr. Harald Bornfleth

Product Manager BESA Research

BESA GmbH

Freihamer Strasse 18

82166 Graefelfing/Germany

http://www.besa.de

E: Harald.Bornfleth at besa.de <mailto:Harald.Bornfleth at besa.de> 

T: +49 89 8980 9968 <tel:%2B49%2089%208980%209968> 

HRB Munich 109956

CEO: Dr. Tobias Scherg, CFO: Theodor Scherg

Director of Research: Dr. Michael Scherg

From: fieldtrip-bounces at science.ru.nl <mailto:fieldtrip-bounces at science.ru.nl>  [mailto:fieldtrip-bounces at science.ru.nl] On Behalf Of Andrew Chang
Sent: Sonntag, 10. Januar 2016 21:18
To: fieldtrip at science.ru.nl <mailto:fieldtrip at science.ru.nl> 
Subject: [FieldTrip] Comparing ICA methods between toolboxes (BESA vs. Fieldtrip)

Dear Fieldtrip users,

I am wondering whether anyone has compared the performances of the ICA methods among M/EEG toolboxes. Especially, how is the performance of ICA in BESA Research 6.0, compared to Fieldtrip?

I have an EEG data set which has been processed in BESA, including removing artifact using ICA. I am wondering whether I can trust it, or I should reprocess it in Fieldtrip. Any comments will be appreciated!

Happy new year!

Best,

Andrew

-- 

Andrew Chang
Ph.D. Candidate
Auditory Development Lab

Department of Psychology, Neuroscience & Behaviour
McMaster University

http://changa5.wordpress.com/

_______________________________________________
fieldtrip mailing list
fieldtrip at donders.ru.nl <mailto:fieldtrip at donders.ru.nl> 
http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

-- 

Andrew Chang
Ph.D. Candidate
Auditory Development Lab

Department of Psychology, Neuroscience & Behaviour
McMaster University

http://changa5.wordpress.com/

-- 

Andrew Chang
Ph.D. Candidate
Auditory Development Lab

Department of Psychology, Neuroscience & Behaviour
McMaster University

http://changa5.wordpress.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160203/2b0500c6/attachment-0001.html>