evoked field amplitude

Thu Jun 3 22:04:29 CEST 2010

I disagree somewhat with the idea that we don't know how to reliably go from sensor to source space. For datasets with large SNR sensory and motor evoked responses, constrained inverse solutions can be quite reliable and correspond very well in some cases with fMRI (whether evoked responses and fMRI "should" correspond to each other is a different question). The literature has many examples of this type of inverse.

Some potential pros, depending on the validity of your source solution and your experiment:

 1.  Reduction of the number of comparisons made (e.g., 275 MEG sensors in the CTF system at NIMH can be reduced to several ECDs in some simple sensory experiments).
 2.  Independence of the source strength from the sensor-brain distance (signal amplitudes in MEG sensors in particular are highly dependent on how far away the head is from the sensory array and this can vary tremendously from subject to subject).
 3.  Higher signal to noise ratio, depending on the approach used. Some spatial filtering approaches such as SSP, beamforming, etc. can substantially reduce the impact of noise from non-brain source such as magnetic dental work artifacts in MEG, gamma-band range muscle activity in EEG, etc.

Some potential cons:

 1.  Invalid source models can nullify any potential advantage.
 2.  In cognitive experiments, some potential sources may not have very large activity relative to the sensory evoked fields/potentials you measure, and depending on the preprocessing and inverse approaches, the source model may be biased towards the larger SNR responses.
 3.  Source modeling can be computationally and/or financially expensive, particularly when highly accurate individual anatomical MRI segmentations are needed to construct the conductor model.

Best,

Don
__________________
Don Rojas, Ph.D.
Director, Magnetoencephalography Laboratory
University of Colorado Denver

On 6/3/10 11:02 AM, "Stanley Klein" <sklein at BERKELEY.EDU> wrote:

Beth, you ask about going to source space.
It's my understanding that people don't yet know how to reliably go from sensor space to source space. The problem is that neighboring cortical areas are often wired together and will fire together and since these neighboring areas will likely have different folding patterns and different time functions the inverse problem of going to sources has problems with both EEG and MEG because of how close these neighbors are to each other. There are two cases where there seem to be solutions, one is occipital cortex where V1, V2, V3, V3a have retinotopic organization that can be identified with fMRI/MRI. The idea here is that with tiny patches in a dartboard layout one can overwhelm the above ambiguity by making use of the topographic layout of cortex. Another area is in somatosensory and motor areas where the homunculus mapping also allows neighboring areas to be separated. Luckily fMRI can also identifying other regions that could be used in helping one go from sensors to sources. However, the above manipulations aren't easily carried out. In addition, there are many researchers who believe that ICA can help do the job.

In summary it is healthy to be skeptical of some of the claims of success in going from sensors to sources.
Stan

On Thu, Jun 3, 2010 at 9:00 AM, Belluscio, Beth (NIH/NINDS) [E] <BelluscB at ninds.nih.gov> wrote:
I want to calculate the strength of the evoked fields in response to a series of stimuli.  Initially, I tried doing this with the averaged data at sensors demonstrating an evoked field following the stimulus.  However, I noticed that during the course of the response, the dipole of the source seemed to shift slightly, and that this dipole was not consistently oriented following successive stimuli.  So I was concerned that by measuring only the response at the sensor, I would obtain a false sense of change in the underlying response.
  I decided to try looking at the response with the data converted into the corresponding planar gradient (I use a CTF MEG).  However, I was unable to find a way to obtain values of the field within an ROI with this data set.  Does anyone know how to do this?
  Lastly, it seems it might be best to compute the source and then evaluate the strength of the signal from an ROI in source-space.  What are the pros/cons of using sensor space vs. source space to measure the amplitude of a response?  Is there a convention within the field for the methodology for doing this?

Beth Belluscio, MD-PhD
Clinical Fellow
Human Motor Control Section
National Institute of  Neurological Disorders and Stroke
301-402-3495

----------------------------------

The aim of this list is to facilitate the discussion between users of the FieldTrip toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis.

http://listserv.surfnet.nl/archives/fieldtrip.html

http://www.ru.nl/fcdonders/fieldtrip/

----------------------------------

The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis.

  http://listserv.surfnet.nl/archives/fieldtrip.html

  http://www.ru.nl/fcdonders/fieldtrip/

----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20100603/7bceabfb/attachment-0002.html>