[FieldTrip] Inverse-modelling requirements

Matt Gerhold matt.gerhold at gmail.com
Tue Sep 20 09:17:33 CEST 2016

MEG Mavens:

I am looking to perform a source-level analysis on some EEG event-related
data. I would be very grateful if you can assist me in understanding some
of the methods in your toolbox and also some of the requirements in terms
of the experimental protocols if one envisages performing source-level
analysis. I have reviewed the tutorials on your website and viewed a number
of video lectures from you institute. I have one or two points I would like
to clear-up and one or two questions that require answers.

>From the available information that I have reviewed, it is recommended that
one have at least the following items: i. hi-res EEG/MEG datasets, ii.
polhemus measurement data, and iii. MRI data for each of the participants
within the study. Having these items enables one to compute the necessary
models to source-localise the EEG/MEG sensor-space data. What I would like
to know is how far one can stretch the boundaries of these requirements and
still produce publishable scientific outcomes: what items are indispensable
to the source localisation methodology?

There are many examples of researchers using standard MRI templates, but
how reliable are analytical outcomes in such instances? Does using a
standard MRI image for all participants really produce useful scientific
outcomes, especially in clinical populations wherein cortical structural
changes are well-documented? There is a fair amount of structural variation
within the cortex across healthy individuals; surely, a single standard MRI
scan would lead to erroneous localisation in some instances?

In terms of electro/magnetic field data: what is the minimum requirement in
terms of how many electrodes are needed (spatial sampling across the scalp)
in order to perform subsequent source-localisation via inverse modelling?
Can one justify using the method(s) in instances of sparse spatial sampling
(32-channels) and expect acceptable scientific outcomes?

If one uses generic sensor/head-model co-registration in the absence of
polhemus data, does this lead to analytical outcomes that are accepted by
yourselves? What are the standards currently being set within the journals;
being mavens in the field, what would you recommend?

I appreciate that most people will embark on the analysis and build
understanding along the way; however, I would like to gain some clarity
before embarking on this analytical journey.

Many thanks in advance.

Kind Regards,

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160920/7178aa13/attachment-0001.html>

More information about the fieldtrip mailing list