Differential EEG

Doktor DynaSoar targeting at OMCL.mil
Thu Mar 4 22:44:13 EST 2004


Editing some more, because, as you admit, some of it is a different
topic.

On Thu, 04 Mar 2004 18:22:03 GMT, "kpaulc" <kpaulc at earthlink.net>
wrote:

} The problem that I'm actually working
} on in this thread is that of rendering the
} neural Topology,

Please explain exactly what you mean by that.

}  graphically, by work-
} ing back from EEG data, which is def-
} initely doable, but which has remained
} undone [unattempted?] - probably be-
} cause the EEG-analysis methodologies
} that are commonly used have been
} handed-down from Professor to Stu-
} dent without any Thought being given
} to rendering data graphically - probably
} be-cause, as 'dynasoar' has pointed-out,
} the methodologies date back to the
} to the Inventor of the EEG, before com-
} puter graphics existed, and probably be-
} cause the demands of developing Ex-
} pertise in the previously-existing meth-
} odology are such that they coerse the
} Student away from thinking graphically.

One problem you haven't confronted is the topography, and by that I
mean sulci and gyrii, of the cortex. EEG picks up sources
perpendicular to the surface. The less perpendicular, the less
sensitive. Sources parallel with the surface don't get seen by the
closest electrodes, and barely at all by distant ones which may be in
the proper attitude.

Seeing it all requires simultaneous MEG, which "sees" perpendicular to
EEG. No great shakes, methodologically. It's doable.

I can, and have, merged simultaneous EEG and MEG data, then overlayed
that on EEG data taken simultaneous with fMRI (MRI plays hell with the
MEG SQUIDs). The electrode placement was digitized per subject and
placed on their anatomical MRI. The results, taken as current source
density, were played back using the real anatomical cortical surface
as the "screen", and presented sample by sample according to the 500
Hz digitization rate. I tested all-pairs coherence. Had I used 3 or
more point Laplacian, I would have done what you've described,
imrpoved upon by use the the subject's real anatomical structure.
Laplacian was available -- I just didn't use it.

None of this software was written in house. It's all commercially
available.




More information about the Neur-sci mailing list