[Eeglablist] baseline removal when running 2 ICAs and frequency of interest finding

[ida miokovic]

Dear Tarik,

thank you very much for your answer.

Preprocessing of my data before 1st ICA consists of highpass filter and
Cleanline line noise removal. Eye and muscle artifacts I hope to detect
after 1st ICA when I would remove them (not my removing components, but
recognizing bad trials). I am not a medical student (electrical engineering
is my area), so I don't know how to recognize bad data just looking at it,
which is one of the suggested techniques of preprocessing.

> Note that there are many choices on how to drop bad epochs when doing
your step 4.
Yes, I have seen that there are 4-5 methods in this step and all of them
requireing some parameters to be entered of course.
- Are there any starting values that you suggest for them with which I can
start (each method is using std.dev as a main parameter)? I As a result, I
know I shouldn't reject more that approx 10% of the data...Makoto suggested
to start with std.dev 8...

I have two more question - I read Arno's suggestion regarding referencing
to the average

"Applying an average reference transformation before applying ICA should
not change the ICA decomposition (since it is a linear operation). However,
we have experienced better quality decompositions if we run ICA prior to
applying average reference, so we advise to apply average-reference
transformations after running ICA (the re-referencing will be also applied
to the ICA weigth matrix)"

- so I suppose that this would be my step No.6? After baseline correction
(on the whole epochs) I should to the re-referencing of the data and in the
same time it will re-reference ICA weigth matrix?
- In case I run ICA on continous data that consists of parts of the
recordings where the subject performed task, is the mentioned procedure the
same? Of course then there would be no step 1 (epoching) and is there
something that has to be done instead of baseline removal, since I wouldn't
have epochs?

Kind regards,

Ida


On Tue, Mar 5, 2013 at 11:45 PM, Tarik S Bel-Bahar
<tarikbelbahar at gmail.com>wrote:

> Hi Ida,
>
> Your steps 1 to 4 seem fine, as long as you are feeding cleaned data to
> the "ICA 1stTime" step.
> After step 4, then you can follow basic steps to get ERP, single-trial,
> frequency, and dipole info on the extra-clean data,
> as described in the eeglab tutorial, and the many articles that use eeglab
> and ICA for their analysis (you can find all of these on Google Scholar).
> Note that there are many choices on how to drop bad epochs when doing your
> step 4.
>
> If you have not done the eeglab tutorial, it is recommended that you do
> it, rather than using your fresh data to figure things out.
> Once you figure out all basic functions and steps with sample data, then I
> suggest pushing your data through some well-decided steps.
>
> Regarding the "baseline removal" based on the Groppe article, you should
> look more closely at those articles to determine exactly
> what you are doing and why you are doing it. you may also want to contact
> the author directly so you have good clarity.
>
> Good luck with your process and looking forward to hearing of your success.
>
> Best wishes,
> Tarik
>
>
> On Sun, Mar 3, 2013 at 5:29 AM, ida miokovic <ida.miokovic at gmail.com>wrote:
>
>>  Dear eeglablist,
>>
>> After reading through many q&a here on eeglab list, I found the
>> recommended literature of Groppe et al regarding baseline removal and went
>> through it.
>>
>> My initial plan was to run ICA on the epoched data 1st time, then to
>> remove bad trials (not the bad components) and then to run ICA second time
>> after which I will hopefully obtain better decomposition. Going through
>> q&a, it wasn't clear to me what is the procedure when ICA is applied 2
>> times. This is short description of the steps as I understand them at the
>> moment:
>> 1. data epoching (without any baseline removal)
>> 2. ICA 1st time
>> 3. bad trials removal
>> 4. ICA 2nd time
>> 5. baseline correction on the whole epochs(?)
>>
>> And also one more question, if I am interested into finding a certain
>> frequency presence in channels and ICs, is Tools --> Component spectra and
>> maps good option? In the field of "Freq.to analyze" I would put the one of
>> my interest and obtain the components (for example 5 of them) and the
>> channel with the highest power of frequency of my interest. I have to learn
>> more about dipfit, since I think I will actually need the source of the
>> signal of the certain frequency localization...
>>
>> Thank you very much in advance,
>>
>> Regards,
>>
>> Ida
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to
>> eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to
>> eeglablist-request at sccn.ucsd.edu
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130306/c4a4e249/attachment.html>