[Eeglablist] dipfit uses average reference?

[David Groppe]

On Wed, 22 Nov 2006, Dien, Joseph wrote:

> Although other references like mean mastoid have the advantages of being
> easy to derive and being readily compared across studies, they make no
> sense from the biophysical perspective.  The problem is that they
> arbitrarily define the reference point as being zero voltage when in
> fact there is no inactive site on the head.  The result is potentially
> severe warping of the solution.  So all source localization algorithms
> that I am aware of use average reference instead, which has the
> advantage of being more plausible from a biophysical perspective.  The
> major weakness of average reference for source solutions is that it
> requires a complete sampling of the voltages over the head to work
> perfectly and, of course, we normally do not sample the underside of the
> head.  I believe that source solution algorithms tend to use an approach
> comparable to Junghofer's PARE correction which uses interpolation to
> estimate the voltages on the bottom of the head.  Insofar as source
> packages like !
>  BESA have been reported to produce plausible results even with
> vertically oriented ERP components like the auditory N1, it appears that
> these estimates work well enough, although more needs to be done on this
> topic.  You can read my 1998 paper for more about average reference
> issues and how it compares to other referencing schemes.  Regarding
> "tends to highlight shallow sources over deep sources", you're thinking
> of current source density (CSD) which is something entirely different.

Thanks for the clarifying note Joe.  The reason why I said that the
average reference tends to highlight deep sources over shallow sources is
based on my experience applying the average reference to some 64 channel
P300 data.  Because the P300 is broadly distributed and positive at almost
all of my electrodes, the average reference severely dampens it relative to
focal components like the N170.  I imagine that the same would be true of
a deep (e.g., Ant. Med. Temporal source) that projects broadly and with
the same polarity to most of my electrodes.  As you said, if I had a more
complete sampling of voltages over the head, this shouldn't be a problem.
  -David


P.S. Wolfgang Skrandies has also noted that average reference acts as a
high-pass spatial filter when applied to moderate density EEG data:

Skrandies, W. (2005) Brain mapping of visual evoked 
activity: Topographical and functional components.  Acta Neurologica 
Taiwanica.

 
> Anyway, bottom line is that I would not recommend trying to disable the
> average reference solution in Dipfit.
> 
> Joe Dien
> University of Kansas
> 
> 
> -----Original Message-----
> From: eeglablist-bounces at sccn.ucsd.edu on behalf of arno
> Sent: Wed 11/22/2006 7:18 PM
> To: David Groppe
> Cc: eeglablist at sccn.ucsd.edu; Robert Oostenveld
> Subject: Re: [Eeglablist] dipfit uses average reference?
>  
> David Groppe wrote:
> > I'm using Dipfit 2 to localize independent components using the spherical 
> > head model.  Apparently the software requires the data to use the average 
> > reference.  Why is this?
> Source localization assumes that the data is average reference (I think 
> it is because no current should get in or out). I do not think it is 
> really an option not to use average reference. Robert might have more 
> insight about that.
> 
> Arno
> > From what I understand, the average reference 
> > tends to highlight shallow sources over deep sources (e.g., medial 
> > temporal), as deep sources have broad, relatively uniform scalp 
> > topographies.  If we're interested in possibly extracting deep sources, is 
> > there a way we can get Dipfit 2 to NOT use the average reference?
> >    much thanks,
> >      -David Groppe
> >
> > dgroppe at cogsci.ucsd.edu
> >   
> _______________________________________________
> eeglablist mailing list eeglablist 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
> 
> 
> _______________________________________________
> eeglablist mailing list eeglablist 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
> 

--