[Eeglablist] ICA on continuous data - eye artifact component

[ida miokovic]

Dear Makoto,

so the only way to obtain the component that have ERPimage that looks more
like one of eye artifact is to remove bad portions from the channels first
and then run ICA again? I will consider that, since I already recognize ICs
like those attached to previous mail as an eye artifact, perhaps it is not
necessary to go through the procedure again..although those bad portions
surely affect other components...

Thanks

Ida


On Sat, Apr 6, 2013 at 3:12 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>wrote:

> Dear Ida,
>
> > Is it possible to remove bad portion only from one IC
>
> No, you have to discard the data from ALL channels. This is because ICA
> learns data across channels and not across time.
>
> Makoto
>
>
> 2013/4/3 ida miokovic <ida.miokovic at gmail.com>
>
>> Dear Makoto,
>>
>> thank you very much for help. So what you are suggesting is that I should
>> manually remove bad portions from IC1 (from scroll data window?) and after
>> doing that, I should get ERPimage that looks more like EOG? Is it possible
>> to remove bad portion only from one IC because if I remember correctly (at
>> the moment I'm in my office, where I do not have EEGLAB) when removing data
>> manually, I am removing data from all components at the particular time
>> period? If my assumption is wrong, what are your suggestions for bad
>> portions removal (data set is continuous)?
>>
>> Cheers,
>>
>> Ida
>>
>>
>> On Wed, Apr 3, 2013 at 3:41 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>wrote:
>>
>>> Dear Ida,
>>>
>>> I saw your plot. This is most likely EOG judging from the topo map, and
>>> contains a few bad portions of the data judging from the ERPimage. Remove
>>> these bad portions and your IC1 ERPimage should look better.
>>>
>>> Makoto
>>>
>>>
>>> 2013/4/2 ida miokovic <ida.miokovic at gmail.com>
>>>
>>>> Dear Makoto,
>>>>
>>>> Within IC1 I see horizontal line in the lower half of the figure,
>>>> meaning this component "is happening" only during one trial (right?), which
>>>> is strange to me since everything else about this component (location,
>>>> activity power spectrum and plot of the component in time domain) is
>>>> showing that it is most likely eye artifact...
>>>>
>>>> In mentioned diagram, I would expect it to be more distributed through
>>>> trials and to look more like:
>>>> http://sccn.ucsd.edu/wiki/File:I94component3_properties.jpg
>>>>
>>>> One data set of mine consists of one subject performing one task (for
>>>> example for 15 secs) than there was a break time (10 secs which I manually
>>>> removed from he data set) and than again task performance etc. What is
>>>> considered in this case to be a trial? Here I have some continuous data
>>>> with markers marking the moments when task is performed and some boundary
>>>> markers when unuseful data (break time) was rejected.
>>>>
>>>> All the best
>>>>
>>>> Ida
>>>>
>>>>
>>>> On Tue, Apr 2, 2013 at 3:27 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>wrote:
>>>>
>>>>> Dear Ida,
>>>>>
>>>>> >Third mentioned diagram is shown with dimensions: 200 trials and 810
>>>>> ms on y and x axis respectfully (image is in the attachment) - why exactly
>>>>> these values are taken? Could I change those?
>>>>>
>>>>> These are chosen arbitrarily. This plot is called ERPimage, and as you
>>>>> can see in its naming, the plot is to show trial-by-trial event-related
>>>>> phase and amplitude structure. Since your data are continuous, the function
>>>>> determined a good number to chop up your data to make them aligned in the
>>>>> box.
>>>>>
>>>>> If you see a 'column' (or 'slope') in the ERPimage that means there
>>>>> was phase-locking event. Do you see it?
>>>>>
>>>>> Makoto
>>>>>
>>>>> 2013/3/30 ida miokovic <ida.miokovic at gmail.com>
>>>>>
>>>>>> Dear list,
>>>>>>
>>>>>> I run ICA on continuous data and now am interested into removal of
>>>>>> the components that represent eye and muscle artifacts (and some other if
>>>>>> recognized).
>>>>>>  When I go through Tools --> Reject data using ICA --> Reject
>>>>>> components by maps there is possibility for me to see Activity power
>>>>>> spectrum, 2D location of component presence and time - trials diagram.
>>>>>> Third mentioned diagram is shown with dimensions: 200 trials and 810 ms on
>>>>>> y and x axis respectfully (image is in the attachment) - why exactly these
>>>>>> values are taken? Could I change those?
>>>>>>
>>>>>>  Since it is IC1 that is shown, I expect it to be most likely an eye
>>>>>> artifact, but I am confused about the look of the third diagram - it does
>>>>>> not look like eye artifact should  and blinks are not visible as shown in
>>>>>> EEGLAB tutorial about ICA. Am I missing somenthing? maybe it is because of
>>>>>> only 800 ms of the dataset on x axis?
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>> 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
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> 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/20130407/08b86312/attachment.html>