[Eeglablist] phase shift and ICA

[Makoto Miyakoshi]

Dear Baris,

I give you the simple answer.
Let's say IC1 and IC2 have regular oscillation like this

IC1 010101010101010101
IC2 001001001001001001

And if your channels are

Ch1 = 0.5xIC1 + 0.5xIC2

Before you discard any IC, your Ch1 should be

Ch1 011011011011011011

If you discard IC1,

Ch1 = 0xIC1 + 0.5xIC2

so your Ch1 will be

Ch1 001001001001001001

that is exactly same as IC2 in terms of phase.

This is exactly what happens to the phase.

Makoto

2013/4/5 Baris Demiral <demiral.007 at gmail.com>

>  Actually the question can be asked in another way:
>
> What happens to phase information when the ICA based artifact rejection is
> used, such that artifactual ICs are taken out, and the data is
> reconstructed back again with the rest of the ICs? Why would I prefer ICA
> correction rather than regression based correction?
>
> Best,
> Baris
>
>
> On 04/04/2013 11:08 AM, Makoto Miyakoshi wrote:
>
> Dear Karlo,
>
>  ICA is to generate spatial filter (EEG.icaweight*EEG.icasphere) and IC
> activities. Scalp channel EEG remains unaffected.
>
>  >Indeed, i like to know: how ICA may affect the  coherency between
> channel_1 (alpha-band) and channel_2 (beta-band)?
>
>  ICA does not change anything, it just generates parallel data.
>
>  Makoto
>
>  2013/4/3 karlo gonzales <thats_karlo at yahoo.com>
>
>>  Thanks for your prompt reply. and my apology, if my question was not
>> clear enough.
>>  My concern is not about phases between channels and ICs.   Indeed, i
>> like to know: how ICA may affect the  coherency between channel_1
>> (alpha-band) and channel_2 (beta-band)? does ICa will shift phase of
>> alpha or beta?
>>
>>  Thanks again,
>> Karlo
>>
>>
>>
>>      ------------------------------
>> *From:* Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> *To:* karlo gonzales <thats_karlo at yahoo.com>
>> *Cc:* EEGLAB List <eeglablist at sccn.ucsd.edu>
>> *Sent:* Wednesday, April 3, 2013 9:36:51 PM
>> *Subject:* Re: phase shift and ICA
>>
>>  Dear Dr. Gonzales,
>>
>>  I suppose you mean you compare phases between channels and ICs. It's
>> not shift that makes difference between scalp channel ICs and EEGs. Let me
>> explain what it is. Let's say we have only 2 channels and 2 ICs. The
>> relation of scalp channel EEG and ICs are like this:
>>
>>  Channel 1 = 0.3 X IC1 + 0.7 X IC2
>> Channel 2 = 0.7 x  IC1 + 0.3 x IC2
>>
>>  Here I determined 0.3 and 0.7 arbitrarily but actually these are
>> determined by ICA. Anyway, when you compare Channel 1 with IC1 in the above
>> example, they are dissimilar to each other NOT because of phase shift
>> (unless IC1 and IC2 are similar, which is hard to imagine since they are
>> supposed to be temporally independent).
>>
>>  If the decomposition is like
>>
>>  Channel 1 = 0.99999 X IC1 + 0.00001 X IC2
>> Channel 2 = 0.00001 x  IC1 + 0.99999 x IC2
>>
>>  Then your Channel 1 and IC1 should be very similar.
>>
>>  Makoto
>>
>>
>>
>> 2013/4/3 karlo gonzales <thats_karlo at yahoo.com>
>>
>>    Dear Dr. *Miyakoshi,*
>> *
>> *
>> *I am using  EEGLAB and ICA tools to clean EEG data. i am recording  EEG
>> from a 64 channel cap (NEUROSCAN), with  a fixed reference electrode, which
>> is  located between CZ and CPZ. *
>> *i do not re-reference my data for further analysis.** So, i just
>> wonder, if applying ICA, would shift the phase of signals (specially low
>> frequency)? is there anyway to check it? *
>> *
>> *
>> *many many thanks for all your help and support,*
>> *Karlo*
>> *
>> *
>>
>>
>>
>>
>>
>>
>>
>>
>>  --
>>  Makoto Miyakoshi
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>>
>>
>
>
>  --
> Makoto Miyakoshi
> Swartz Center for Computational Neuroscience
> Institute for Neural Computation, University of California San Diego
>
>
> _______________________________________________
> 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
>
>
>


-- 
Makoto Miyakoshi
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/20130405/73382dca/attachment.html>