[Eeglablist] ASR parameters for adults vs. kids/clinical populations

[Delorme, Arnaud]

Dear Anna,

The value of 20 for ASR is the default in clean_rawdata 2.1.

If the data is too noisy, you might want to have ASR correct bad data portions instead of rejecting them - clean_rawdata 2.1 plugin will reject data by default (but ASR is used to detect bad portions of data that would need correction and reject these instead). The reason, ASR does not correct data by default is that ASR correction is mostly useful when you have very little data or when you are running realtime applications (which ASR was originally designed for). For offline data processing, it is better to be conservative (the less mortification to the raw data, the better) and simply reject the few portions of data which are labeled as bad by ASR.

Best wishes,

Arno

> On Jan 21, 2020, at 2:26 PM, Anna Kasdan <avkasdan at gmail.com> wrote:
> 
> Hi all,
> 
> Thanks for this information. I am re-running things with the new ASR
> parameter recommendations, as I had used some of the old recommendations
> before.
> 
> And in fact Gedeon we have the same paradigm in infants, children, adults,
> and different clinical populations so maybe at some point I could delve
> into this systematically!!
> 
> Anna
> 
> 
> On Tue, Jan 21, 2020 at 1:46 PM Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
> wrote:
> 
>> Dear Anna,
>> 
>> ＞I am using ASR to clean EEG data and am wondering if anyone recommends
>> using different parameter values in the ASR function for adult data (i.e.
>> relatively clean) vs. kid/clinical data (relatively messy).
>> 
>> For the recommended parameters, see these papers.
>> 
>> https://sccn.ucsd.edu/wiki/Artifact_Subspace_Reconstruction_(ASR)#Reference_.2809.2F11.2F2019_update.29
>> 
>> ＞Is it possible also for ASR to "overclean" data from a dataset that is
>> originally pretty
>> good?
>> 
>> Generally speaking, ASR tend to cut a lot. The point of the very lax
>> criteria (>> 10) recommended in the abovelinked papers is to adjust the
>> amount of cutting.
>> 
>> Basically, for standard data use 20 all the time. For messy data, use
>> 10-20. For my chronic tic patient's data (subjects were ticing and blinking
>> during EEG recording), I used 6-8. See Loo et al. (2019) NeuroImage
>> Clinical.
>> 
>> Makoto
>> 
>> 
>> 
>> On Tue, Jan 21, 2020 at 12:49 AM Anna Kasdan <avkasdan at gmail.com> wrote:
>> 
>>> Hi all,
>>> 
>>> I am using ASR to clean EEG data and am wondering if anyone recommends
>>> using different parameter values in the ASR function for adult data (i.e.
>>> relatively clean) vs. kid/clinical data (relatively messy). Is it
>> possible
>>> also for ASR to "overclean" data from a dataset that is originally pretty
>>> good?
>>> 
>>> Thank you!
>>> 
>>> Anna Kasdan
>>> _______________________________________________
>>> 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
>>> 
>> _______________________________________________
>> 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
> _______________________________________________
> 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