[FieldTrip] Resting State Data Source Reconstruction - covriance matrix in beamformer, and "avg" field in eLORETA

[Haiteng Jiang]

>
> Dear  Roey,
>

         I don't have answer  to your  question but I have  some thoughts.

         Firstly , you can cut your EEG resting state data into trials (2s
per trial) . Since the dominant  frequency of  resting state is  alpha ,
you can build  frequency domain beamformer 'DICS' with (cfg.normalize='yes')
to normalize the lead field , which will avoid bias toward the center of
the head (http://fieldtrip.fcdonders.nl/tutorial/beamformer).  After that ,
you can use  virtual channel (all voxels) to extract the time series (
http://fieldtrip.fcdonders.nl/tutorial/shared/virtual_sensors).   However,
I am not sure whether it makes sense or not . Maybe the experts can comment
on it.

                                                                  Best,
                                                                  Haiteng



>
>
> Message: 2
> Date: Tue, 15 Apr 2014 17:03:00 +0300
> From: Roey Schurr <roeysc at gmail.com>
> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
> Subject: [FieldTrip] Resting State Data Source Reconstruction -
>         covriance matrix in beamformer, and "avg" field in eLORETA
> Message-ID:
>         <CAHm4wZA4KuZeG7O0V0xL2G6jM4Cb9usY-Cj7_Fh9c=
> zYfDVYjg at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Fieldtrippers,
>
>
> First of all, a big thank you to all members of the community ? it is a
> wonderful place to learn in and share ideas, thoughts and solutions to
> common problems we all face.
>
>
> I am writing you since I fear I might be missing something trivial: I am
> trying to compute an inverse solution to to resting-state EEG data. Two
> question arise, depending on the method I am trying to use:
>
>
> 1)      Using a beamformer technique such as LCMV, a covariance matrix is
> needed. However, since the data is not time-locked (like ERP), the
> covariance matrix cannot be easily computed over the averaged trials. Can
> it be estimated using different resting-state trials?
>
> 2)      Using eLORETA as the source reconstruction method (since my goal is
> a time-domain source reconstruction), it seems the the "avg" field is neeed
> in the data. Once again, no such average is available, since these are not
> ERP trials.
>
> Would you advise me to revise the code and try to fit it to non-ERP data,
> or is there some easier way I'm missing?
>
>
> Thank you all,
>
> Best,
>
> roey
>
>
>

-- 
Haiteng Jiang
PhD candidate
Neuronal Oscillations Group
Donders Institute for Brain, Cognition and Behaviour
Centre for Cognitive Neuroimaging
Radboud University Nijmegen

Visiting address
Room 2.32
Donders Centre for Cognitive Neuroimaging
Kapittelweg 29
6525 EN  Nijmegen
the Netherlands

 Tel.: +31 (0)243668291
 Web:  https://sites.google.com/site/haitengjiang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140418/e4791b70/attachment-0001.html>