[Eeglablist] Help on ICA components in data analysis

Wed Mar 27 09:38:16 PDT 2013

Dear Imali,

>But the most posterior electrodes did not co register properly
(warp_1.jpg).

Oh your channel coregistration is very bad. This should be properly done,
otherwise you'll never get a valid result.

Adjust parameters manually and don't use auto-warping or anything if it
returns error. You have to have 10-5 nomenclatures to make it work etc.
In doing so, your channels never get completely overlapped but that's fine.
Try to find the parameters that minimizes the error (visually inspected, of
course) as a whole.

>I know you might be very busy with the work and you valuable time is
really appreciated.

I'm on the beach and sending it from my iPhone. I'm lying.

Makoto

2013/3/26 IMALI THANUJA HETTIARACHCHI <ith at deakin.edu.au>

>  Dear Makoto,
>
> Yes the issue has been in coregistering the electrodes on the head. And it
> is not completely still solved.
>
> I did a manual coregistration by setting the Yaw to -pi/2 and then by
> auto-selecting the channels to match the electrode locations. But the most
> posterior electrodes did not co register properly (warp_1.jpg).
>
> Then I tried to pair the these electrodes (I,CB1,CB2,CB1",CB2") manually
> with the reference montage eg. I to Iz, then I got the error,
>
> ??? Error using ==> minus
> Matrix dimensions must agree.
>
> Error in ==> coregister>warp_chans at 654
>     dpre = mean(sqrt(sum((elec1.pnt(ind1,:) - elec2.pnt(ind2,:)).^2, 2)));
>
> Error in ==> coregister at 197
>                     [ tmp dat.transform ] = warp_chans(dat.elec1,
> dat.elec2,
>                     tmpelec2.label(clist2), 'traditional');
>
> ??? Error using ==> waitfor
> Error while evaluating uicontrol Callback
>
> I tried with the MNI head model, but even with the Spherical model this
> was the same.
>
> And when I tried the DIPFIT with this head model, the component 1 dipole
> was still not anywhere near the visual area(dipole_location.jpg).
>
> I cannot figure out what I can do about this, can you please advice?
>
> Component 1 , to me is a clearly a visual activation and I expect to see
> the dipoles in the occipital area. Am I not getting this because of the
> coregistering problem?
>
> In order to comment on the connectivity analysis I should first have my
> source localisations and the sources selected properly, which I am having
> no luck with. Do you have any suggestions what I can try for this?
>
> I know you might be very busy with the work and you valuable time is
> really appreciated.
>
> Thank you very much.
>
> Kind regards
> Imali
>
>  ------------------------------
> *From:* Makoto Miyakoshi [mmiyakoshi at ucsd.edu]
> *Sent:* Saturday, 23 March 2013 11:15 AM
> *To:* IMALI THANUJA HETTIARACHCHI
>
> *Subject:* Re: Help on ICA components in data analysis
>
>  no worries. good luck.
>
> 2013/3/22 IMALI THANUJA HETTIARACHCHI <ith at deakin.edu.au>
>
>>  Thank you so much Makoto, this helped me so much, I will go further
>> with the analysis and keep you updated too.
>>
>> Regards
>> Imali
>>  ------------------------------
>> *From:* Makoto Miyakoshi [mmiyakoshi at ucsd.edu]
>> *Sent:* Saturday, 23 March 2013 7:31 AM
>> *To:* IMALI THANUJA HETTIARACHCHI
>> *Cc:* EEGLAB List
>> *Subject:* Re: Help on ICA components in data analysis
>>
>>   Dear Imali,
>>
>>  No, your dipoles are not coregistered correctly. Redo the head modeling.
>>
>>  >Their selected activations are in visual, mid central and left motor
>> cortex. Therefore I was looking at selecting components 1, 8, 10, 16,
>> 17,18, 22, 23,25 for the connectivity analysis.
>>
>> This seems fine.
>> Always remember that the more ICs you include, the more difficult it is
>> for SIFT to model the network well. If you don't see good results in
>> modeling, try to reduce the number of ICs (though 9 ICs should be fine).
>>
>>  Makoto
>>
>> 2013/3/21 IMALI THANUJA HETTIARACHCHI <ith at deakin.edu.au>
>>
>>>  Dear Makoto,****
>>>
>>> ** **
>>>
>>> Things have been held up on our own data collection as there has been
>>> amendments for the ethics approval.  ****
>>>
>>> ** **
>>>
>>> Therefore I decided to go ahead with some available real data for my
>>> connectivity analysis. I downloaded and had a good exploring on the
>>>   animal/distractor task data available on
>>> http://sccn.ucsd.edu/~arno/fam2data/publicly_available_EEG_data.html . *
>>> ***
>>>
>>> ** **
>>>
>>> What I need to do is to identify and extract the ICA activations (source
>>> waveforms) of the components that look like neural activity for this visual
>>> categorisation task.  I initially looked at the ICA maps for the subject
>>> ‘cba’ on the ‘animal’ task (cbaanimal.set) and the the component maps are
>>> as follows;****
>>>
>>> ****
>>>
>>> ** **
>>>
>>> However when I looked at the dipole locations also available with the
>>> data set, the 1st dipole corresponding to the 1st ICA component was in
>>> the attached figure. This was a bit confusing as the component 1 clearly is
>>> a bilateral occipital dipole.  ****
>>>
>>> ** **
>>>
>>> **1.)    **Do you see any reason for this? And/or what can I do to fix
>>> this?****
>>>
>>> ** **
>>>
>>> **2.)    **Also accordin to the first published work on this data set
>>>
>>> ****
>>>
>>> Delorme, A., Rousselet, G., Mace, M., Fabre-Thorpe M. Interaction of
>>> Bottom-up and Top-down processing in the fast visual analysis of natural
>>> scenes*. Cognitive Brain Research*, 103-113. Author's PDF<http://www.sccn.ucsd.edu/%7Earno/mypapers/DelormeCBR2003.pdf>
>>> ****
>>>
>>> ** **
>>>
>>> Their selected activations are in visual, mid central and left motor
>>> cortex. Therefore I was looking at selecting components 1, 8, 10, 16,
>>> 17,18, 22, 23,25 for the connectivity analysis.****
>>>
>>> ** **
>>>
>>> Do you think these are reasonable?****
>>>
>>> ** **
>>>
>>> I really appreciate any guidance and help that you can give for my
>>> analysis.  Thank you very much for your valuable time. ****
>>>
>>> ** **
>>>
>>> Kind regards****
>>>
>>> Imali****
>>>
>>> * *
>>>
>>
>>
>>
>>  --
>> Makoto Miyakoshi
>> JSPS Postdoctral Fellow for Research Abroad
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>
>
>
>  --
> Makoto Miyakoshi
> JSPS Postdoctral Fellow for Research Abroad
> Swartz Center for Computational Neuroscience
> Institute for Neural Computation, University of California San Diego
>

-- 
Makoto Miyakoshi
JSPS Postdoctral Fellow for Research Abroad
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130327/a0dbf98c/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 36294 bytes
Desc: not available
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130327/a0dbf98c/attachment.png>