[Eeglablist] ICA, re-referencing, interpolation

[Makoto Miyakoshi]

Dear Jumana,

AMICA does not do any drastic improvement compared with infomax. So if you
saw bad results in infomax (runica), AMICA would return the bad results as
well.

> Is it nexeeeary for me to change to amica?

If you say you have time and effort resource to do it, yes it is worth. But
again, you should not expect any drastic changes in results. Remember GIGO:
garbage in, garbage out.

> Also, if you do interpolate ICA then how many channels do you remove to
prevent rank deficiency and how does this depend on the number of channels
interpolated?

First of all, you never interpolate ICA activations. You can interpolate
channels because it is less important information than IC sources, and the
purpose of channel interpolation is ONLY FOR spatially-unbiased average
referencing.

>From UA49 to Newark,

Makoto



On Fri, Jan 13, 2017 at 1:28 AM, Ahmad, Jumana <jumana.ahmad at kcl.ac.uk>
wrote:

> Dear Makoto,
> I also noticed EEG lab have changed to AMICA but I've already run ICA on
> 100 participants using the previously recommended runica. I was never rank
> deficient (i average ref after ICA).
> I do it have component corruption. Is it nexeeeary for me to change to
> amica? It would mean redoing work.
> Also, if you do interpolate ICA then how many channels do you remove to
> prevent rank deficiency and how does this depend on the number of channels
> interpolated?
> Best wishes,
> Jumana
> ------------------------------------------
> Jumana Ahmad
> Post-Doctoral Research Worker in Cognitive Neuroscience
> EU-AIMS Longitudinal European Autism Project (LEAP) & SynaG Study
> Room M1.26.Department of Forensic and Neurodevelopmental Sciences (PO 23)
> | Institute of Psychiatry, Psychology & Neuroscience | King’s College
> London | 16 De Crespigny Park | London SE5 8AF
>
> Phone: 0207 848 5359| Email: jumana.ahmad at kcl.ac.uk | Website:
> www.eu-aims.eu | Facebook: www.facebook.com/euaims
>
> ------------------------------
> *From:* Ahmad, Jumana
> *Sent:* 13 January 2017 11:22:18
> *To:* mmiyakoshi at ucsd.edu
>
> *Cc:* eeglablist at sccn.ucsd.edu
> *Subject:* Re: [Eeglablist] ICA, re-referencing, interpolation
>
>
> Hi Makoto,
> I think your code does, but I am following a mix of your pipeline when
> appropriate but also reordering this it to keep in line with what my
> colleagues in the same project are doing, and they interpolate after. So I
> am trying to follow those steps. I am hoping that if I delete the ICA
> information and then interpolate and rereference the raw data,that this
> should be fine? I will check carefully but when I view the raw data after
> rereferencing, the blinks are still removed.
> Best wishes,
> Jumana
> ------------------------------------------
> Jumana Ahmad
> Post-Doctoral Research Worker in Cognitive Neuroscience
> EU-AIMS Longitudinal European Autism Project (LEAP) & SynaG Study
> Room M1.26.Department of Forensic and Neurodevelopmental Sciences (PO 23)
> | Institute of Psychiatry, Psychology & Neuroscience | King’s College
> London | 16 De Crespigny Park | London SE5 8AF
>
> Phone: 0207 848 5359| Email: jumana.ahmad at kcl.ac.uk | Website:
> www.eu-aims.eu | Facebook: www.facebook.com/euaims
>
> ------------------------------
> *From:* Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
> *Sent:* 13 January 2017 10:02:59
> *To:* Ahmad, Jumana
> *Cc:* eeglablist at sccn.ucsd.edu
> *Subject:* Re: [Eeglablist] ICA, re-referencing, interpolation
>
> Dear Jumana,
>
> > I am trying to replicate a pipeline very precisely and this pipeline
> interpolates after ICA.
>
> Oops, is that true? But when I checked the latest description now, I found
> that they do NOT change channels after AMICA. Press 'refresh' button to
> update the page if you haven't done so.
>
> > If possible I would like to interpolate after, as I only need to work
> with the channel data after the ICA.
>
> If that's the goal, then I think it's fine. But test it carefully before
> you press the final button! If you are not confident, I recommend you use
> dead copy of my code and use the channel information in the end (assuming
> that my latest code does NOT interpolate channels after ICA).
>
> Makoto
>
>
>
> On Thu, Jan 12, 2017 at 11:31 PM, Ahmad, Jumana <jumana.ahmad at kcl.ac.uk>
> wrote:
>
>>
>> Hi Makoto,
>> I am trying to replicate a pipeline very precisely and this pipeline
>> interpolates after ICA. I only wanted to remove the artefacts from the data
>> and then convert back to channel data. If I delete the eye blink components
>> and then convert back to raw data, -and then delete all ICA info, such as
>> the weights and activations etc, is it then safe for me to interpolate
>> using the channel data (which now has blinks removed)?
>>
>> If I really have to avoid the situation, then how do you get around rank
>> issues of interpolating before? I know you delete channels but how do you
>> know how many to delete in relation to the number of channels interpolated.
>> If possible I would like to interpolate after, as I only need to work with
>> the channel data after the ICA.
>>
>> Thank you again for your advice! Very much appreciated.
>> Jumana
>> ------------------------------------------
>> Jumana Ahmad
>> Post-Doctoral Research Worker in Cognitive Neuroscience
>> EU-AIMS Longitudinal European Autism Project (LEAP) & SynaG Study
>> Room M1.26.Department of Forensic and Neurodevelopmental Sciences (PO 23)
>> | Institute of Psychiatry, Psychology & Neuroscience | King’s College
>> London | 16 De Crespigny Park | London SE5 8AF
>>
>> Phone: 0207 848 5359| Email: jumana.ahmad at kcl.ac.uk | Website:
>> www.eu-aims.eu
>> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.eu-aims.eu&data=01%7C01%7Cjumana.ahmad%40kcl.ac.uk%7C89d274d92ca94825ded208d43b8ab8d2%7C8370cf1416f34c16b83c724071654356%7C0&sdata=b2bm%2BLN%2BqlU%2Ff%2BfqHuP1cDUBtGGS8VSHrYwdKzl5d8A%3D&reserved=0>
>> | Facebook: www.facebook.com/euaims
>> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.facebook.com%2Feuaims&data=01%7C01%7Cjumana.ahmad%40kcl.ac.uk%7C89d274d92ca94825ded208d43b8ab8d2%7C8370cf1416f34c16b83c724071654356%7C0&sdata=G%2BWt3IuifCY5wZcspxBMYq1li3tZFcTBhSynPUzpTCc%3D&reserved=0>
>>
>> ------------------------------
>> *From:* Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> *Sent:* 13 January 2017 08:05:04
>> *To:* Ahmad, Jumana
>> *Cc:* eeglablist at sccn.ucsd.edu
>> *Subject:* Re: [Eeglablist] ICA, re-referencing, interpolation
>>
>> Dear Jumana,
>>
>> I encountered that kind of situation many times, which is why I thought
>> it would be a nice idea to show the exact steps in the wiki page. Now you
>> see my point!
>>
>> Editing channels after ICA is DANGEROUS. There are even some phenomena
>> that confused me to death which I still don't understand (and do not
>> remember well any more). Some of EEGLAB's check function can detect the
>> change of channel number difference and inconsistency with EEG.ica-items,
>> but not all the operations are not supported, particualy any behavior
>> outside GUI (but the worst mysterious case I experienced was replicated
>> even using GUI, so I guess it was like I was performing bug finding in
>> EEGLAB, as if several test players plays an alpha version of a computer
>> game trying to do the weirdest things to find bugs.)
>>
>> So, my advice is that DO NOT EDIT CHANNELS AFTER ICA. Once the mixing
>> matrix is confused and how it happened, you are done. Just stay away from
>> that situation.
>>
>> I appreciate though you have been doing exploration! I'm glad to see that
>> at least other guys than I experience the same issues.
>>
>> From Lufthansa455 to Frankfurt,
>>
>> Makoto
>>
>>
>>
>> On Wed, Jan 4, 2017 at 9:06 AM, Ahmad, Jumana <jumana.ahmad at kcl.ac.uk>
>> wrote:
>>
>>> Dear EEGLab,
>>>
>>>
>>>
>>> I do not include noisy channels in ICA which I only do to remove eye
>>> related components, and specifically want to interpolate then after ICA for
>>> rank reasons. I have now done this to over 100 datasets.
>>>
>>> However, once I interpolate and then re-reference my data to the
>>> average, I get the error: Error: some channels not used for ICA
>>> decomposition are used for rereferencing
>>>
>>>        the ICA decomposition has been removed
>>>
>>>
>>>
>>> However, I purposefully wanted to interpolate these after ICA, and then
>>> use them for the average.
>>>
>>> Is my eye-blink removal via ICA now affected and undone?
>>>
>>> I would appreciate any advice.
>>>
>>> Best wishes,
>>>
>>> Jumana
>>>
>>>
>>>
>>> *------------------------------------------*
>>>
>>> *Jumana Ahmad*
>>>
>>> Post-Doctoral Research Worker in Cognitive Neuroscience
>>>
>>> *EU-AIMS Longitudinal European Autism Project (LEAP) & SynaG Study*
>>>
>>> Room M1.09. Department of Forensic and Neurodevelopmental Sciences (PO
>>> 23) | Institute of Psychiatry, Psychology & Neuroscience | King’s College
>>> London | 16 De Crespigny Park | London SE5 8AF
>>>
>>>
>>>
>>> *Phone:* 0207848 0260| *Email:* jumana.ahmad at kcl.ac.uk
>>> <antonia.sanjose at kcl.ac.uk> | *Website:* www.eu-aims.eu
>>> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.eu-aims.eu%2F&data=01%7C01%7Cjumana.ahmad%40kcl.ac.uk%7C283380d9389643d9653708d43b7a38bc%7C8370cf1416f34c16b83c724071654356%7C0&sdata=NdhNvFWrTx1tn28VuA31t4ZIiyIMuEoGJ2%2FR5XnRKaI%3D&reserved=0>
>>>  | *Facebook:* www.facebook.com/euaims
>>> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.facebook.com%2Feuaims&data=01%7C01%7Cjumana.ahmad%40kcl.ac.uk%7C283380d9389643d9653708d43b7a38bc%7C8370cf1416f34c16b83c724071654356%7C0&sdata=%2FCH%2FgGc5kXxLAiHkczUoYIw4SUc1pFQvcYYgWAra3Lc%3D&reserved=0>
>>>
>>>
>>>
>>> ************************************************************
>>> ***************************
>>>
>>> ************************************************************
>>> ***************************
>>>
>>> *We are currently looking for volunteers with mild intellectual
>>> disability to be part of our exciting and world-leading European project
>>> into brain development and social behaviour. Please, do get in touch if you
>>> know of anyone who may be interested in taking part. *
>>>
>>> ************************************************************
>>> ***************************
>>>
>>> ************************************************************
>>> ***************************
>>>
>>>
>>>
>>> _______________________________________________
>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fsccn.ucsd.edu%2Feeglab%2Feeglabmail.html&data=01%7C01%7Cjumana.ahmad%40kcl.ac.uk%7C283380d9389643d9653708d43b7a38bc%7C8370cf1416f34c16b83c724071654356%7C0&sdata=Smt9AFupZWONzIqfUxI6kozsJPaRQQJLcPy1csF9oHs%3D&reserved=0>
>>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.uc
>>> sd.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
>>
>
>
>
> --
> 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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20170121/4cfb2b35/attachment.html>