[Eeglablist] eeglablist Digest, Vol 144, Issue 14

Mon Oct 24 15:50:49 PDT 2016

Dear James,

Thank you for sharing your experience. I want to know more about it if you
don't mind.

> It frequently led to seemingly highly-correlated channel time-series
data. Then the resulting ICA patterns were not very interpretable:
ironically, every component seemed to basically represent a single
electrode, with a very focal peak in the topo map, and very little neural
activity in the ERP image.

First of all, results from ICA (I assume we see it using 'reject component
by maps' Somehow this is the best visualization tool to confirm the
multiple aspects of ICs at a glance) are sorted by the order of variance.
This means that if you have 64ch for example, the top 15-20 ICs would
explain more than 95% of variance of the data, and the rest is basically
noise. So my question is, do you see the single-electrode topographic data
in these top 15-20 ICs, or later 'noise' components?

Another thing: ASR learns the covariance matrix from the cleanest part of
the data. Channel data are naturally highly corrected because of volume
conductance, normally you should see high correlation across channels.
However, observing single-channel scalp maps means something opposite: they
are not correlated with others, by definition! It seems that these two are
mixed in your report. Are you sure that the channel-correlation you saw are
pathological? How do you judge so?

Diagnosing the problem without seeing data is hard, I should just imagine...

Makoto

On Thu, Oct 20, 2016 at 12:21 PM, James Jones-Rounds <jj324 at cornell.edu>
wrote:

> Hi Vishal, Makoto, and all,
>
> I have also had a similar issue with running ASR. I tried several
> parameter combinations for the Burst Criterion (values of 3 and 5), and
> Window Criterion (values of 0.1 and 0.25), and each combination of these.
> It frequently led to seemingly highly-correlated channel time-series data.
> Then the resulting ICA patterns were not very interpretable: ironically,
> every component seemed to basically represent a single electrode, with a
> very focal peak in the topo map, and very little neural activity in the ERP
> image.
>
> For some datasets, however, the ASR pre-ICA cleaning step really cleared
> up the data and revealed a number of neural components that manual cleaning
> did not reveal. I haven't determined yet what the critical difference is
> that affects the ASR performance..
>
> James
>
> On Thu, Oct 20, 2016 at 3:00 PM, <eeglablist-request at sccn.ucsd.edu> wrote:
>
>> Send eeglablist mailing list submissions to
>>         eeglablist at sccn.ucsd.edu
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>>         https://sccn.ucsd.edu/mailman/listinfo/eeglablist
>> or, via email, send a message with subject or body 'help' to
>>         eeglablist-request at sccn.ucsd.edu
>>
>> You can reach the person managing the list at
>>         eeglablist-owner at sccn.ucsd.edu
>>
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of eeglablist digest..."
>>
>> Today's Topics:
>>
>>    1. Re: Highly Correlated channels after applying ASR
>>       (Makoto Miyakoshi)
>>    2. Re: Interpreting Brodmann Area and Anatomical location of
>>       Domain in MPT analysis (Makoto Miyakoshi)
>>    3. Re: Simultaneous recording of EEG and Robot motion
>>       (Makoto Miyakoshi)
>>
>>
>> ---------- Forwarded message ----------
>> From: Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> To: Vishal Vijayakumar <vijay059 at umn.edu>
>> Cc: EEGLAB List <eeglablist at sccn.ucsd.edu>
>> Date: Wed, 19 Oct 2016 20:00:22 -0700
>> Subject: Re: [Eeglablist] Highly Correlated channels after applying ASR
>> Dear Vishal,
>>
>> What do you see when you perform average reference? How many channels do
>> you have by the way? Also, what types of cap are you using? If EGI, check
>> if they were bridged by any chance...
>>
>> Makoto
>>
>> On Mon, Oct 17, 2016 at 3:08 PM, Vishal Vijayakumar <vijay059 at umn.edu>
>> wrote:
>>
>>> Hello,
>>>
>>> Apologies if you've answered this question already, but I observe that
>>> data across my channels look highly correlated after applying ASR to the
>>> filtered EEG data. My project involves analyzing tonic thermal pain, so I
>>> apply ASR to 4 minute segments of the data. Because of the high
>>> correlation, my ICA sphering matrix is quite large in magnitude (minimum
>>> value of 1800) for many subjects. I also disable any channel-based
>>> rejection because I'd like to concatenate data across subjects for group
>>> ICA. My current problem persists even during single subject analysis.
>>>
>>>
>>> I'm not sure if I'm doing something wrong, and I'd appreciate any
>>> insight you can provide!
>>>
>>> Thanks,
>>> Vishal
>>>
>>> _______________________________________________
>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>> 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
>>
>>
>> ---------- Forwarded message ----------
>> From: Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> To: Mohammed Jarjees <m.jarjees.1 at research.gla.ac.uk>
>> Cc: "eeglablist at sccn.ucsd.edu" <eeglablist at sccn.ucsd.edu>
>> Date: Wed, 19 Oct 2016 20:15:49 -0700
>> Subject: Re: [Eeglablist] Interpreting Brodmann Area and Anatomical
>> location of Domain in MPT analysis
>> Dear Mohammed,
>>
>> Sorry for belated response.
>> I asked the question to Nima and obtained the following response.
>>
>> 1. The data come from two different atlases.
>> 2. Any area less than 0.5 contribution is not shown (wihch is why they
>> don't sum up to 1).
>> 3. For Brodmann area (in Talairach), it uses the closest cortical voxel.
>> Basal ganglia is removed from this.
>>
>> Probably 1 and 2 explains the difference.
>>
>> Makoto
>>
>> On Sun, Sep 18, 2016 at 1:57 AM, Mohammed Jarjees <
>> m.jarjees.1 at research.gla.ac.uk> wrote:
>>
>>> Dear all,
>>>
>>> I have used Measure Projection Toolbox (MPT) to analyse  motor imagery
>>> EEG data.  In some domains I get Brodmann areas contributing and
>>> corresponding anatomical areas which are not covering these Brodmann areas
>>> at all (example below). If they are complementary, then why BA adds up to
>>> 91% and anatomical areas up to  83%?
>>>
>>>
>>>
>>> *Brodmann areas*
>>>
>>> *Probability*
>>>
>>> BA 25
>>>
>>> BA 23
>>>
>>> BA 34
>>>
>>> BA 29
>>>
>>> BA 28
>>>
>>> 0.47
>>>
>>> 0.16
>>>
>>> 0.16
>>>
>>> 0.06
>>>
>>> 0.06
>>>
>>>
>>>
>>> *Total Probability*
>>>
>>> *0.91*
>>>
>>>
>>>
>>> *Anatomical areas*
>>>
>>> *Probabilities*
>>>
>>> L Caudate
>>>
>>> Brainstem
>>>
>>> R Caudate
>>>
>>> L Putamen
>>>
>>>
>>>
>>> *Total Probability*
>>>
>>>
>>>
>>> 0.29
>>>
>>> 0.23
>>>
>>> 0.17
>>>
>>> 0.14
>>>
>>>
>>>
>>> *0.83*
>>>
>>>
>>>
>>>
>>>
>>> I understand that anatomical areas can be wider than BAs, but shouldn't
>>> it anatomical areas cover BAs?
>>>
>>> How should I interpret this?
>>>
>>>
>>>
>>> Best Regards
>>>
>>> Mohammed
>>>
>>>
>>>
>>> _______________________________________________
>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>> 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
>>
>>
>> ---------- Forwarded message ----------
>> From: Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> To: Sanjaya Vipula <sanjayavipula at gmail.com>
>> Cc: EEGLAB List <eeglablist at sccn.ucsd.edu>
>> Date: Wed, 19 Oct 2016 20:45:54 -0700
>> Subject: Re: [Eeglablist] Simultaneous recording of EEG and Robot motion
>> Dear Sanjaya,
>>
>> Right, maybe post-hoc signal processing can treat it well. At least, you
>> can optimize the hardware so that the noise can be easier to be handled
>> with.
>>
>> Makoto
>>
>> On Tue, Oct 18, 2016 at 11:15 PM, Sanjaya Vipula <sanjayavipula at gmail.com
>> > wrote:
>>
>>> Dear Makoto,
>>>
>>> Exactly as you anticipated, the behavior is really different depending
>>> on the conductivity.
>>>
>>> However the major problem is when the motor is operating the noise level
>>> is too high. I am going to try to identify a exact noise band related to
>>> the motor movement and to filter it as some of the members have suggested.
>>>
>>> Thank you for your comment.
>>>
>>> regards,
>>>
>>> Sanjaya
>>>
>>> On 18 October 2016 at 16:56, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>>> wrote:
>>>
>>>> Dear Sanjaya,
>>>>
>>>> Hmm that's a difficult question.
>>>> Probably you can test it step by step. For example, while touching the
>>>> robot with a rubber hand vs. bare hand, or using wood stick vs. rubber
>>>> stick vs. conductive stick, or touch, close but no touch, no touch far,
>>>> etc. I'm sure the artifact level is highest when subject's body touches the
>>>> robot. Could there be any non-conductive material between a subject and and
>>>> the robot to prevent the artifact from contaminating EEG?
>>>>
>>>> Makoto
>>>>
>>>> On Tue, Oct 11, 2016 at 12:29 AM, Sanjaya Vipula <
>>>> sanjayavipula at gmail.com> wrote:
>>>>
>>>>> Dear All,
>>>>>
>>>>> I need some advice on a problem I face during EEG recording.
>>>>>
>>>>> I am trying to record EEG simultaneously with an exoskeleton movement.
>>>>> The subject wear the exoskeleton during the experiments.
>>>>>
>>>>> When the exoskeleton is not operating normal EEG signals can be
>>>>> observed from subject. However when the exoskeleton is operating a huge
>>>>> noise can observed in the EEG recordings.
>>>>>
>>>>> I tried separating the power source of the EEG amplifier and the
>>>>> robot. The noise is probably due to the motor operation. How the effect
>>>>> from the motor operation can be suppressed in the EEG signal?
>>>>>
>>>>> Thank you very much in advance for any comments and tips.
>>>>>
>>>>> regards,
>>>>>
>>>>> Sanjaya
>>>>>
>>>>> --
>>>>> <sanjayavipula at gmail.com>Sanjaya Vipula Bandara,
>>>>> B.Sc. Mech.Eng.(SL), MPhil.Biorobotics(SL), AMIESL
>>>>> PhD Student,
>>>>> Department of Mechanical Engineering,
>>>>> Kyushu University,
>>>>> Fukuoka,
>>>>> Japan
>>>>>
>>>>> _______________________________________________
>>>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>>>> 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
>>>>
>>>
>>>
>>>
>>> --
>>> <sanjayavipula at gmail.com>Sanjaya Vipula Bandara,
>>> B.Sc. Mech.Eng.(SL), MPhil.Biorobotics(SL), AMIESL
>>> PhD Student,
>>> Department of Mechanical Engineering,
>>> Kyushu University,
>>> Fukuoka,
>>> Japan
>>>
>>
>>
>>
>> --
>> Makoto Miyakoshi
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>> _______________________________________________
>> eeglablist mailing list eeglablist at sccn.ucsd.edu
>> Eeglablist page: http://www.sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to eeglablist-unsub at sccn.ucsd.edu
>> To switch to non-digest mode, send an empty email to
>> eeglablist-nodigest at sccn.ucsd.edu
>>
>
>
>
> --
> James Jones-Rounds
> Laboratory Manager
> Human Development EEG and Psychophysiology (HEP) Laboratory,
> Department of Human Development,
> --------------------------------------------
> Cornell University | Ithaca, NY
> 607-255-9883
> eeg at cornell.edu
>
> _______________________________________________
> 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/20161024/d6805697/attachment.html>