[Eeglablist] Urgent help

Thu Jan 26 15:42:47 PST 2017

Dear Andria and Arno,

Yes Arno is right. I confused FFT with Wavelet Transform.
My previous comment,

> In your case though, you want to compare whether using the whole epoch
baseline makes the data look more reasonable or not. If you do this, then
you are evaluating the deviation from the mean across all recording time.

applies to the case of power calculated by Wavelet Transform.

> Baseline is only subtracted (in log or linear space) when performing
time-frequency decompositions (newtimef function).

This is equivalent to dividing all datapoints by pre-stimulus mean, so this
is normalization.

Makoto

On Thu, Jan 26, 2017 at 2:50 PM, Arnaud Delorme <arno at ucsd.edu> wrote:

> There is no baseline “normalization” when computing the data or ICA
> component spectrum using spectopo. Baseline is only subtracted (in log or
> linear space) when performing time-frequency decompositions (newtimef
> function).
>
> Arno
>
> On Jan 26, 2017, at 2:43 PM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu> wrote:
>
> Dear Andria,
>
> > 1- In the case of mine, where there is no markers, and then no baseline
> normalization, do you think produces power specturm with dB (produced by
> spectopo) can lead to problems is correct? or using microvolts squared
> (uV2) should correct? or both dB and microvolts squared are acceptable?
>
>
> All are correct. That being said, the problem of using uV^2 is that the
> value tend to be ridiculously large. Convert to dB makes the data more
> intuitively understandable. Therefore I personally prefer to converting to
> dB. In your case though, you want to compare whether using the whole epoch
> baseline makes the data look more reasonable or not. If you do this, then
> you are evaluating the deviation from the mean across all recording time.
>
>
> > 2- Does "spectopo" function apply baseline normalization by default from
> EEGLAB GUI?
>
> Yes. To turn it off, you need to use an optional input.
>
> Makoto
>
>
>
> On Thu, Jan 26, 2017 at 1:19 AM, Andria Lan <andrialan108 at gmail.com>
> wrote:
>
>> Dear Makoto,
>>
>> I have an issue that is related to our previous discussion.
>>
>> As you know that my data don't have any markers, hence, my analysis will
>> focuses mainly on ERSPs. I'll display the following points and then ask my
>> questions.
>>
>> In one of the EEGLAB discussions, *according to Arno: **spectrum
>> returned in EEGLAB is in unit dB* which is *10*log10(uV^2/Hz)*.
>> Everything is good till now. In addition, according to some references:
>>
>> *dBtf = 10*logtf*(activitytf / baselinef)*.
>>
>>
>> Now, please focus at this point:
>>
>> In this last dB formula, you can see that the existence of both *dB* and*
>> baseline (baseline normalization)*. This means that, existence of dB is
>> required the existence of baseline normalization. Beside, baseline
>> normalization can be implemented in the case of having markers in order to
>> place the baseline before the stimulus onset. In sum, with dB and baseline
>> normalization are located with dataset that have markers where power
>> spectrum unit in this case is dB (that is mentioned by Arno up) which I
>> believe produced by spectopo.
>>
>>
>>
>> Consequently, my two questions:
>>
>>
>> 1- In the case of mine, where there is no markers, and then no baseline
>> normalization, do you think produces power specturm with dB (produced by
>> spectopo) can lead to problems is correct? or using microvolts squared
>> (uV2) should correct? or both dB and microvolts squared are acceptable?
>>
>>
>> 2- Does "spectopo" function apply baseline normalization by default from
>> EEGLAB GUI?
>>
>>
>> Thanks for your aptionance and assistance.
>>
>>
>> Andria
>>
>>
>> On Thu, Jan 26, 2017 at 12:54 AM, Andria Lan <andrialan108 at gmail.com>
>> wrote:
>>
>>> Dear Makoto,
>>>
>>> Thanks a ton. Without you and your help, I really don't know what to do.
>>>
>>> Wish you all the best and the success as well.
>>>
>>> Andria
>>>
>>> On Wed, Jan 25, 2017 at 3:04 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>>> wrote:
>>>
>>>> Dear Andria,
>>>>
>>>> If you don't have event markers, you cannot apply event-related
>>>> potential analysis. All you can use is basically spectra in EEGLAB STUDY.
>>>>
>>>> > 1- For first subject, how did you recommend saving the recorded data
>>>> (during the experiment) to be used later at preprocessing stage and STUDY
>>>> option?
>>>>
>>>> Save as a single .set file. Again, if you don't have events, that's the
>>>> only option for you.
>>>>
>>>> > 2- However, do you recommend *ALWAYS* using **only** one file to
>>>> save the data for each subject (all conditions) with both ERP and ERSP
>>>> analysis?
>>>>
>>>> Yes, I recommend that. Historically, EEGLAB supported separate .set
>>>> files for different conditions, but now STUDY.design can handle them, and
>>>> the old method tend to have compatibility problem so less stable.
>>>>
>>>> Makoto
>>>>
>>>>
>>>> On Tue, Jan 17, 2017 at 6:12 PM, Andria Lan <andrialan108 at gmail.com>
>>>> wrote:
>>>>
>>>>> Dear Makoto,
>>>>>
>>>>> Thanks a lot for your prompt reply and the useful link. However, I
>>>>> don't have any events (markers) for participants' recorded data. I have
>>>>> only the signal resulting from the task.
>>>>>
>>>>> Here is the scenario:
>>>>>
>>>>> For the sake of clarity (for now), I have:
>>>>>
>>>>> a) 30 subjects
>>>>> b) 2 conditions.
>>>>> c) 10 visual task trials for each condition.
>>>>> d) subjects only need to see the trials.
>>>>> e) not button pressing-->no ERP analysis (analysis only based
>>>>> "time-frequency" or "ERSP")
>>>>>
>>>>> My questions:
>>>>>
>>>>> 1- For first subject, how did you recommend saving the recorded data
>>>>> (during the experiment) to be used later at preprocessing stage and STUDY
>>>>> option?
>>>>>
>>>>> 2- However, do you recommend *ALWAYS* using **only** one file to save
>>>>> the data for each subject (all conditions) with both ERP and ERSP analysis?
>>>>>
>>>>>
>>>>> 
>>>>>  STUDY for loading the data.png
>>>>> <https://drive.google.com/file/d/0B430Bz2U6pH-N0J1TnJITk5rblk/view?usp=drive_web>
>>>>> 
>>>>> Thank you.
>>>>> Andria
>>>>>
>>>>> On Fri, Jan 13, 2017 at 3:30 PM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu
>>>>> > wrote:
>>>>>
>>>>>> Dear Andria,
>>>>>>
>>>>>> If you follow our recommended preprocessing steps, you don't need to
>>>>>> average any data before STUDY does it on it own.
>>>>>>
>>>>>> > is it enough (and correct way) loading the data files (for each
>>>>>> subject) one-by-one under each condition using the “STUDY” option of EEGLAB?
>>>>>>
>>>>>> Do not separate .set into conditions. See this section.
>>>>>> https://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline
>>>>>> #Create_STUDY_.2801.2F05.2F2017_updated.29
>>>>>>
>>>>>> Makoto
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Jan 12, 2017 at 1:58 AM, Andria Lan <andrialan108 at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Dear EEGLAB list,
>>>>>>>
>>>>>>>
>>>>>>> I need your advice about implementing this scenario using EEGLAB
>>>>>>> toolbox.
>>>>>>>
>>>>>>> I have this issue and I need your advice:
>>>>>>>
>>>>>>>
>>>>>>> I performed my experiment on several subjects, the task doesn't
>>>>>>> required any events because it's only about watching several trials where
>>>>>>> each belongs to one specific condition. Hence, I ended up with several data
>>>>>>> files for each subject. In addition, I don’t have any epochs in those
>>>>>>> files, and my aim is performing time-frequency analysis. Now, in order to
>>>>>>> do such analysis using EEGLAB toolbox, is it enough (and correct way)
>>>>>>> loading the data files (for each subject) one-by-one under each condition
>>>>>>> using the “STUDY” option of EEGLAB?
>>>>>>>
>>>>>>>
>>>>>>> Do average the results (files) for each subject is required at this
>>>>>>> stage, or by using the STUDY option this method will be accomplished
>>>>>>> automatically?
>>>>>>>
>>>>>>>
>>>>>>> Any help would be highly appreciated.
>>>>>>>
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Andria
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> 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
> _______________________________________________
> 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/20170126/32905043/attachment.html>