[Eeglablist] LORETA applied to single-trial EEG analyses?

[Makoto Miyakoshi]

Dear Eric,

> This doesn't bode well for the use of dipole fitting via dipfit either in
the localization of EEG regression results, does it (as dipfit also relies
on Maxwell's equations)?

You are right. If the same reasoning applies, you can't fit dipole to scalp
map of regression results.

> I guess the only way to estimate the cortical source of a scalp
regression result would be to run the regression on source-level data such
as independent components or voxelwise activity returned from LORETA (or
something similar)?

You are right.

Makoto

On Thu, Jul 11, 2019 at 9:37 PM Eric Rawls <elrawls at email.uark.edu> wrote:

> Makoto,
> Thanks, I had assumed as much. Of course there is no reason to apply
> physiological electromagnetic constraints to beta weights.
> This doesn't bode well for the use of dipole fitting via dipfit either in
> the localization of EEG regression results, does it (as dipfit also relies
> on Maxwell's equations)?
> I guess the only way to estimate the cortical source of a scalp regression
> result would be to run the regression on source-level data such as
> independent components or voxelwise activity returned from LORETA (or
> something similar)?
> Thanks for the response
> Eric
>
> On Thu, Jul 11, 2019 at 6:49 PM Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
> wrote:
>
>> Dear Eric,
>>
>> > Are these weights acceptable as input to LORETA?
>>
>> Most likely not.
>>
>> According to the site you referenced, LORETA calculates dipoles in brain
>> voxels. A dipole is a electromagnetic property, so the input value should
>> be compatible with that. Thus, if the input data is just linear
>> transformation of a scalp measurement, such as re-referencing, it would be
>> fine. However, if you use non-EEG data as input, it would loose the
>> validity the author claims because the algorithm calculates electric
>> behavior of non-electric property. For example, what is the field theory of
>> the beta weights? Does it follow Maxwell's equation? Why? I guess the
>> assumption of the calculation becomes weird in this way.
>>
>> I could be wrong here, I would appreciate if someone knowledgeable can
>> confirm my explanation.
>>
>> Makoto
>>
>> On Mon, Jul 8, 2019 at 4:51 AM Eric Rawls <elrawls at email.uark.edu> wrote:
>>
>>> Hi list, question for those who have used or are familiar with LORETA.
>>>
>>> LORETA accepts either time- or frequency-domain EEG. However, the
>>> following
>>> resource notes a number of ways input to LORETA could be "cheated"
>>> despite
>>> seeming reasonable inputs.
>>>
>>> https://www.uzh.ch/keyinst/NewLORETA/Misuse/Misuse.htm
>>>
>>> My question relates to the cortical localization of beta weights from
>>> single-trial regression analysis of EEG. This type of analysis
>>> essentially
>>> computes an "erp" using regression rather than averaging.
>>>
>>> Are these weights acceptable as input to LORETA? This approach was taken
>>> in
>>> Fischer & Ullsperger (2013) to localize cortical prediction error
>>> coefficients, but I confess I'm quite unsure whether this is a correct
>>> use
>>> of LORETA.
>>>
>>> What do we think?
>>> Eric Rawls
>>> PhD Candidate
>>> Department of Psychological Sciences
>>> University of Arkansas
>>> _______________________________________________
>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>> To unsubscribe, send an empty email to
>>> eeglablist-unsubscribe at sccn.ucsd.edu
>>> For digest mode, send an email with the subject "set digest mime" to
>>> eeglablist-request at sccn.ucsd.edu
>>>
>>
>>
>> --
>> Makoto Miyakoshi
>> Assistant Project Scientist, Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>

-- 
Makoto Miyakoshi
Assistant Project Scientist, Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego