It is important to weight one's interpretation dipole fitting results from dipfit (or other inverse methods) by how much or little residual scalp map variance the component map has from the scalp projection of the equivalent dipole model - We often ignore independent components whose scalp maps have residual variances above 15-20% (though this is not a hard and fast distinction, and should depend in some way on the number of channels recorded). <br>
<br>ICA gives many components that are highly 'dipolar' (with low residual variances) -- and, in results we will soon submit for publication, gives more such components the more mutual information ICA removes from the data, though the number of such components identified by ICA varies somewhat from dataset to dataset depending, first, on the number of data channels, and then on other (less well defined) 'data quality' factors. Removing stretches of spatially non-stereotyped ('messy') data from the data to be decomposed by ICA will definitely increase the numbers of dipolar components identified in the data.<br>
<br>Scott Makeig<br><br><br><br><div class="gmail_quote">On Fri, Dec 31, 2010 at 6:42 AM, Andrew Dimitrijevic <span dir="ltr"><<a href="mailto:adimitri@uci.edu">adimitri@uci.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<div bgcolor="#ffffff" text="#000000">
Hi Dorothy,<br>
<br>
This is a common problem (at least with auditory evoked potentials).
Some suggestions that worked for me in the past:<br>
<br>
1) use "initial seed points" that are approximately in the auditory
cortex (you can set the initial x,y,z points) before you do the
dipole fit<br>
2) make sure you have a good SNR (general comment for source
analysis, but seems to be extra important for DipFit)<br>
<br>
the reason that you're getting dipoles in the centre of the head is
that it "truely" is the best explanation of the variance, albeit it
doesn't make sense physiologically. <br>
<br>
cheers<br>
<br>
andrew<div><div></div><div class="h5"><br>
<br>
<br>
On 12/30/2010 3:01 AM, Dorothy Bishop wrote:
</div></div><blockquote type="cite"><div><div></div><div class="h5">
<div style="font-family: Tahoma; direction: ltr; color: rgb(0, 0, 0); font-size: 13px;">
<div dir="ltr"><font size="2" face="Tahoma"><font size="2" face="Courier New"><font size="2" face="Courier New"><font size="2" face="Courier New">
<p>I am learning to use Dipfit2 with eeglab9_0_4_5s.</p>
<p><font face="courier new">All works well if I accept
default head model.</font></p>
<p><font face="courier new">I would like, however, to
modify the head model for children.</font></p>
<p> </p>
<p><font face="courier new">I followed the
instructions for doing this that I found in:</font></p>
<p><a href="http://sccn.ucsd.edu/pipermail/eeglablist/2007/001690.html" target="_blank"><font color="#000000">http://sccn.ucsd.edu/pipermail/eeglablist/2007/001690.html</font></a></p>
<p> </p>
<p><font face="courier new">I kept everything the same
as for spherical 4 shell model, except that I
subsituted a new head model file with altered
vol.r values to correspond to 8 yr old child. I
also, on advice from a colleague, made</font></p>
<font size="2" face="Courier New"><font size="2" face="Courier New">
<p>vol.c(3)=.0084</p>
<p> </p>
<p><font face="courier new">Output coordinates
default to MNI when you select custom model
file, and I kept this.</font></p>
<p> </p>
<p><font face="courier new">I then went to Edit
Channel locations to set head radius to be
same as last vol.r value, i.e. 70.3</font></p>
<p> </p>
<p><font face="courier new">I then ran autofit
dipoles , specifying 'fit bilateral dipoles'.</font></p>
<p><font face="courier new">But I end up with
single dipoles in the middle of the head.</font></p>
<p> </p>
<p><font face="courier new">I am not sure if the
problem is with my model specification, or
whether there is a bug, and would be grateful
for advice.</font></p>
<p> </p>
<p><font face="courier new">thanks</font></p>
<p> </p>
</font></font></font></font></font></font></div>
<div dir="ltr"> </div>
<div><font size="2">
<div>Dorothy Bishop, Professor of
Developmental Neuropsychology,<br>
Dept of Experimental Psychology, University of Oxford, OX1
3UD.<br>
tel +44 (0)1865 271369; fax +44 (0)1865 281255;<br>
WEB: <a href="http://www.psy.ox.ac.uk/oscci/" target="_blank">http://www.psy.ox.ac.uk/oscci/</a><br>
Blog: <a href="http://deevybee.blogspot.com/" target="_blank">http://deevybee.blogspot.com/</a></div>
</font></div>
</div>
</div></div><pre><fieldset></fieldset>
_______________________________________________
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><div class="im">
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu" target="_blank">eeglablist-unsubscribe@sccn.ucsd.edu</a>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu" target="_blank">eeglablist-request@sccn.ucsd.edu</a></div></pre>
</blockquote>
</div>
<br>_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.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.edu</a><br></blockquote></div><br><br clear="all"><br>-- <br>Scott Makeig, Research Scientist and Director, Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla CA 92093-0559, <a href="http://sccn.ucsd.edu/%7Escott" target="_blank">http://sccn.ucsd.edu/~scott</a><br>