[Eeglablist] Pop_rejchan with kurtosis returns NaN

[Makoto Miyakoshi]

Dear Adrien,

> Removing bad channels from epochs would be ideal, is there a function
that supports it?

No, that's not available.

> I had been looking for one but only found methods for continuous data.

I see your point. There is Christian Kothe's plugin 'clean_rawdata'
available.
http://sccn.ucsd.edu/wiki/Plugin_list_process
This one cleans data for each 1-sec window AND channel. It's currently the
smartest way to clean the raw EEG as far as I know. Here is Christian's
explanation of how it works.
http://sccn.ucsd.edu/eeglab/plugins/ASR.pdf

Makoto

On Thu, Dec 10, 2015 at 4:51 AM, Adrien Martel <amartel at tcd.ie> wrote:

> Dear Makoto,
>
> thanks very much for the swift reply. I will use the amplitude threshold
> then, I’ve never used the improbability test but will definitely give it a
> look.
>
> Removing bad channels from epochs would be ideal, is there a function that
> supports it? I had been looking for one but only found methods for
> continuous data.
>
> Adrien
>
> On December 10, 2015 at 00:55:39, Makoto Miyakoshi (mmiyakoshi at ucsd.edu)
> wrote:
>
> Dear Adrien,
>
> In calculating kurtosis if you have continuous zeros or near-zero values
> it would fail. Remove those problematic portions before calculating
> kurtosis.
>
> By the way I don't recommend you use kurtosis for cleaning epochs. Use
> simple amplitude threshold at +/-500 or around (i.e. absolutely
> unacceptable), then improbability test with SD==5-8 (i.e. relatively bad).
>
> Makoto
>
> On Tue, Dec 8, 2015 at 7:12 AM, Adrien Martel <amartel at tcd.ie> wrote:
>
>> When using pop_rejchan with kurtosis (normalised) in the results one of
>> the channels is returned with the Measure NaN. Other methods of that
>> function return a value between 0.5-1.
>>
>> What could be the reason for this? This is after importing raw data,
>> filtering (0.5-40Hz with separate lowpass and then highpass) and
>> downsampling (512->256).
>> When I look at the raw data, the channel seems weirdly flat but the
>> values do change. This happens with 2 subjects out of 3 on the same channel
>> (POz). I would just interpolate it after the rest of the analysis but that
>> channel is of particular interest to my study.
>>
>> Any help would be greatly appreciated,
>> Adrien
>>
>>
>>
>> _______________________________________________
>> 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/20151210/554cc232/attachment.html>