[FieldTrip] Pre-whitening/Denoising Elekta/Neuromag Resting State Data

[Sarah]

Hi Jan-Mathijs,

Thank you for your response!

Yes, I agree that using other resting state data may throw away the baby
with the bathwater as you say. I suppose what I was thinking when writing
my message and mentioned to "make use of the resting state recordings" was
potentially using "a part" of the resting state data as a way to estimate
the noise data - that is, preprocessing and filtering the RS data in the
upper most range, ex. 150Hz+ - upper limit Hz, to use as the estimate of
the noise data in the pre-whitening function. (This freq range I'm guessing
may more likely contain less physiologic data and more noise).

I guess maybe this still might slightly throw away the baby with the
bathwater, but nevertheless was just trying to think about solutions to
improve beamforming in cases where it is maybe more necessary
(ex.connectivity/cross-frequency coupling analyses).

Nevertheless, thanks again for your help!
Sarah

On Thu, May 16, 2024 at 5:18 PM Sarah <s83728498 at gmail.com> wrote:

> Hello Fieldtrippers!
>
>
> I’ve recently learned from this wonderful paper by our very own Fieldtrip
> developers “A unified view on beamformers for M/EEG source reconstruction” (
> https://www.sciencedirect.com/science/article/pii/S1053811921010612) that
> - when working with the Elekta/Neuromag system, if SSS is applied through
> the MEGIN software for cleaning (ex. MaxFilter), regularization alone is
> not sufficient for good beam former performance. Instead, one
> recommendation was pre-whitening in combination with using a truncated
> pseudo-inverse.
>
>
> Since I also have MEG data acquired from an Elekta with SSS applied, I am
> interested in applying this pre-whitening step on Fieldtrip, particularly
> to my resting state data. For pre-whitening of this data, one
> recommendation from the paper is using an empty room recording as the
> inputted noise data to the pre-whiten function. However, in my case,
> unfortunately I do not have these to accompany each of my resting state
> recordings.
>
>
> I was wondering if anyone may have any recommendations on how I may still
> follow through with pre-whitening to ensure good beam former performance in
> my case. Are there any potential alternatives to the input noise data to
> the prewhiten function in Fieldtrip suitable for resting state data (ex.
> Something that potentially makes use of the resting state recordings)?
>
>
> If, on the contrary, it seems that the pre-whiten function is not
> something I could use in my case, I would also appreciate any alternate
> functions/methods/steps that I could take to still ensure the unwanted
> effects of the SSS by the MEGIN software are minimized (ie rank deficiency).
>
>
>
> Thanks and I would greatly appreciate any thoughts or suggestions,
>
> Sarah
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240527/81ceecbc/attachment.htm>