[Eeglablist] About ICA

[Makoto Miyakoshi]

Dear Leonardo,

Delorme et al. (2007) NeuroImage technical paper showed that epoch
rejection on IC activations (not components!) rather than channel EEG is
2SD more efficient etc. If you haven't checked out the paper I recommend
you do.

Makoto

On Fri, Dec 18, 2015 at 9:53 PM, Leonardo Lion <llionname at gmail.com> wrote:

> Dear Makoto,
>
> Thank you so much for your helpful advices.
>
> In fact, I didn't get the idea behind copying ICA weights from dataset to
> other, especially that I did this but didn't find any changes in the
> resulting dataset from the copying process.
>
> Any highlights about this point?
>
> Regards,
> Leonardo
> On 19 Dec 2015 02:54, "Makoto Miyakoshi" <mmiyakoshi at ucsd.edu> wrote:
>
>> Dear Leonardo,
>>
>> > 1- As I understood, we need to apply the ICA data because it is clean
>> to the (e.g., raw data that is not clean). Is this correct? I.E., do I need
>> to copy data from data number 7 to raw data (number one?)
>>
>> The recommendation is that you run ICA, then you perform epoch rejection
>> ON ICA-DECOMPOSED TIME SERIES (i.e. IC activation), then run ICA again.
>>
>> > Let's say I performed ICA and now my data reached number 7 in the
>> preprocess stage within eeglab. Now, I'm satisfy with the results of this
>> data after ICA, how can I apply this note on the data Is it necessaay to
>> apply the results (copy it) to the dataset 1(raw data)? if yes, then I need
>> to do the follwing steps:
>>
>> No, save the dataset instead. That's your cleaned data.
>>
>> > 2- What is the next step after this important note? after copying the
>> data, what should we do?
>>
>> Compare ERP/ERSP/ITC/etc across conditions to answer your question.
>>
>> FYI, there is alternative and unofficial wiki page for the preprocessing
>> issue.
>> http://sccn.ucsd.edu/wiki/Makoto's_preprocessing_pipeline
>>
>> Good luck.
>>
>> Makoto
>>
>> On Wed, Dec 16, 2015 at 6:49 AM, Leonardo Lion <llionname at gmail.com>
>> wrote:
>>
>>> Hi EEGLAB list,
>>>
>>> I have quesnito reslted to the impplimentation of ICA, my quesitno
>>> belong to the link
>>>
>>> http://sccn.ucsd.edu/wiki/Chapter_09:_Decomposing_Data_Using_ICA
>>>
>>> where there is this note:
>>>
>>> *Important note:* we believe an optimal strategy is to:
>>>
>>>    1. Run ICA
>>>    2. Reject bad epochs (see the functions we developed to detect
>>>    artifactual epochs and channels, if any, in the tutorial on artifact
>>>    rejection <http://sccn.ucsd.edu/wiki/Chapter_01:_Rejecting_Artifacts>).
>>>    In some cases, we do not hesitate to remove more than 10% of the trials,
>>>    even from 'relatively clean' EEG datasets. We have learned that it is often
>>>    better to run this first ICA composition on very short time windows.
>>>    3. Run ICA a second time on the 'pruned' dataset.
>>>    4. Apply the resulting ICA weights to the same dataset or to longer
>>>    epochs drawn from the same original (continuous or epoched) dataset. For
>>>    instance, to copy ICA weights and sphere information from dataset 1 to 2:
>>>    First, call menu Edit > Dataset info of dataset 2. Then enter
>>>    *ALLEEG(1).icaweights* in the *ICA weight array ...* edit box,
>>>    *ALLEEG(1).icasphere* in the *ICA sphere array ...* edit box, and
>>>    press *OK*.
>>>
>>>
>>> 1- As I understood, we need to apply the ICA data because it is clean to
>>> the (e.g., raw data that is not clean). Is this correct? I.E., do I need to
>>> copy data from data number 7 to raw data (number one?)
>>>
>>> Let's say I performed ICA and now my data reached number 7 in the
>>> preprocess stage within eeglab. Now, I'm satisfy with the results of this
>>> data after ICA, how can I apply this note on the data Is it necessaay
>>> to apply the results (copy it) to the dataset 1(raw data)? if yes, then I
>>> need to do the follwing steps:
>>>
>>> Edit > Dataset info of dataset 1. Then enter *ALLEEG(7).icaweights* in
>>> the *ICA weight array ...* edit box, *ALLEEG(7).icasphere* in the *ICA
>>> sphere array ...* edit box, and press *OK*.
>>>
>>>
>>> 2- What is the next step after this important note? after copying the
>>> data, what should we do?
>>>
>>> Thanks.
>>> Leonardo
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20151221/31a2a1f9/attachment.html>