[FieldTrip] Can the cortical source of any sensor-level metric be localized?

Li Su ls514 at cam.ac.uk
Mon Mar 29 17:16:20 CEST 2021


Hi Ingmar,

My suggestion is that if you are interested in the source, you’d better to do MNE localisation first and run RSA directly in the source space. Here are two papers from my own group applying this method.

Wingfield, C., Su, L. , Liu, X., Zhang, C., Woodland, P., Thwaites, A., Fonteneau, E., Marslen-Wilson, W.D., (2017) Relating Dynamic Brain States to Dynamic Machine States: Human and Machine Solutions to the Speech Recognition Problem, PloS Computational Biology. doi.org/10.1371/journal.pcbi.1005617 <http://doi.org/10.1371/journal.pcbi.1005617>

Su, L. , Fonteneau, E., Marslen-Wilson, W. and Kriegeskorte, N. (2012) Spatiotemporal Searchlight Representational Similarity Analysis in EMEG Source Space, IEEE Xplore, 97-100, doi:10.1109/PRNI.2012.26

Bw,

Li Su


— 
Professor Li Su, PhD
Chair of Neuroimaging
Sheffield Institute for Translational Neuroscience, 
University of Sheffield, 385A Glossop Road, 
Sheffield, S10 2HQ, UK
Alzheimer’s Research UK Senior Research Fellow
Department of Psychiatry, University of Cambridge, 
Level E4, Box 189, Cambridge Biomedical Campus, 
Cambridge, CB2 0SP, UK

> On 29 Mar 2021, at 15:43, Ingmar de Vries <i.e.j.de.vries at gmail.com> wrote:
> 
> Dear Fieldtrip community,
> 
> I have a conceptual question that potentially applies to many possible situations. If it helps, here is my situation: 
> 
> I am currently analyzing MEG data (Neuromag, 306 sensors), using representational similarity analysis (RSA). I have applied a spatial searchlight at the sensor level, which results in RSA (i.e. correlation) values at each of the 306 sensors. This results in an interesting spatial peak of correlation above areas that are expected to be involved. 
> 
> Is it possible to localize the cortical source of this sensor-level peak in correlation? 
> This question can be generalized to any metric, i.e.: Is it possible to localize the cortical source of any sensor-level metric?
> 
> I can't find any literature doing something similar, and the main problem I foresee is in the definition of a head model. That is, standard head models either define electrical conductivity (EEG), or spread of the magnetic field (MEG), but not the spread of an information-based metric (here correlation, but could be decoding accuracy as well). Is there a solution for this?
> 
> The alternative approach (plan B) would be to first apply source reconstruction on the raw MEG signal with a distributed source model (e.g. MNE), and subsequently apply the RSA analyses on the source-reconstructed signal. 
> 
> Thanks for the input and I hope you are all healthy and safe, and able to continue your research in these strange times!
> 
> cheers,
> 
> -- 
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Ingmar de Vries, PhD.
> Postdoc @ CIMeC, University of Trento
> i.e.j.de.vries at gmail.com <mailto:i.e.j.de.vries at gmail.com>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20210329/8a4a7ea8/attachment.htm>


More information about the fieldtrip mailing list