<div dir="ltr">Dear Jonathan,<div><br></div><div>Sorry for belated response. Your questions are legitimate!</div><div><br></div><div>> First, we our montage is not high density – we only have 32 data electrodes.</div><div><br></div><div>It does not matter.</div><div>'Artifact Subspace' in ASR means, if I understand correctly, principal components associated with multiple channels (i.e spatial distribution of PCs). The number of channels or their distributions does not matter.</div><div><br></div><div>> Second, our approach at the moment is to distribute them unevenly to cover particular areas of interest – I will try to attach and image of the montage. I was wondering if people could weigh in on the importance of this to ASR. For instance, might it be desirable to distribute the electrodes evenly?</div><div><br></div><div>If you have a noisy channel and epoch in a scalp regions covered only sparsely, then the 'artifact subspace' will include fewer number of channels. That's it. It does not matter. </div><div><br></div><div>> Second, are there ballpark opinions about whether ASR should be used with only 32 electrodes? Just eyeballing it, the EEG looks reasonable with the “burst” parameter set to 20, but some electrodes end up getting removed, leaving us with even fewer electrodes as inputs to the algorithm.</div><div><br></div><div>The noisy channel removal is performed BEFORE ASR. So if you want to protect channels, you should rather change the parameters for channel rejections. ASR parameter has nothing to do with the final number of channels you get!</div><div><br></div><div>> My worry is that this will result in more lossy corrections since ASR as I understand it relies on correlations between electrodes.<br></div><div><br></div><div>In an extreme case, you can keep all channels and all datapoints while cleaning data... but in this case what happens is that you'll get pretty meaningless data in the places where the chunk of data should be usually rejected by the program.</div><div><br></div><div>From Lufthansa455 to Frankfurt,</div><div><br></div><div>Makoto</div><div><br></div><div><br></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Dec 20, 2016 at 7:30 AM, Jonathan Folstein <span dir="ltr"><<a href="mailto:jonathan.r.folstein@gmail.com" target="_blank">jonathan.r.folstein@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">
<p class="MsoNormal">Dear eeglab, we are considering using artifact subspace
reconstruction for offline artifact correction/rejection in a mobile EEG
experiment. We are using the cognionics dry mobile EEG cap shown in this video </p>
<p class="MsoNormal"> </p>
<p class="MsoNormal"><a href="https://www.youtube.com/watch?v=qYC_3SUxE-M" target="_blank">https://www.youtube.com/watch?<wbr>v=qYC_3SUxE-M</a></p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">...but after having read a bit about ASR, there are some potentially
relevant differences between our setup and other things that have been tried
with ASR, e.g. here...</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal"><a href="https://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline" target="_blank">https://sccn.ucsd.edu/wiki/<wbr>Makoto's_preprocessing_<wbr>pipeline</a></p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">First, we our montage is not high density – we only have 32
data electrodes. Second, our approach at the moment is to distribute them
unevenly to cover particular areas of interest – I will try to attach and image
of the montage.<span> </span>I was wondering if
people could weigh<span> </span>in on the importance
of this to ASR.<span> </span>For instance, might it
be desirable to distribute the electrodes evenly? Second, are there ballpark
opinions about whether ASR should be used with only 32 electrodes?<span> </span>Just eyeballing it, the EEG looks reasonable
with the “burst” parameter set to 20, but some electrodes end up getting
removed, leaving us with even fewer electrodes as inputs to the algorithm. My
worry is that this will result in more lossy corrections since ASR as I
understand it relies on correlations between electrodes.</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">thanks, and happy holidays</p><span class="gmail-HOEnZb"><font color="#888888">
<p class="MsoNormal"> </p>
<p class="MsoNormal">Jonathan</p>
<br clear="all"><br>-- <br><div class="gmail-m_-144805229175998088gmail_signature"><div dir="ltr"><div><div>Jonathan Folstein<br>Assistant Professor<br>Department of Psychology<br>Florida State University <br><font size="2"><span style="font-family:arial,helvetica,sans-serif">1107 W. Call St. <br>Tallahassee, FL 32306-4301</span></font>
</div>office: <a href="tel:(850)%20645-0654" value="+18506450654" target="_blank">850-645-0654</a><br></div><div>fax: <a href="tel:(850)%20644-7739" value="+18506447739" target="_blank">850-644-7739</a></div><a href="mailto:folstein@psy.fsu.edu" target="_blank">folstein@psy.fsu.edu</a><br></div></div>
</font></span></div>
<br>______________________________<wbr>_________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" rel="noreferrer" target="_blank">http://sccn.ucsd.edu/eeglab/<wbr>eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.<wbr>ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.<wbr>edu</a><br></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature"><div dir="ltr">Makoto Miyakoshi<br>Swartz Center for Computational Neuroscience<br>Institute for Neural Computation, University of California San Diego<br></div></div>
</div></div>