Beamformer

[MARUYAMA Masaki INSERM]

Dear Fieldtrip users,

 

 

Please let me ask two questions about beamformer.

 

(1) Beamformer for different-type sensors

I have two different types of sensors: planer gradiometers and magnetometers. I would like to know if beamformers should compute the different-type signals separately. If so, after beamformer sourceanalysis, is it appropriate to integrate two NAIs (neural activity indexes) in such as (NAI_mag + NAI_grad)/2 or NAI_mag x NAI_grad? I would appreciate any comments/ advices on this issue. 

 

(2) Common filters in LCMV-beamformer

Although the Fieldtrip website does NOT tell that common filters are available in LCMV, I'm wondering if it is limited to cases of single trial beamformer. According to the discussion of Drs Floris and Robert on 2009 Feb 2, the current version of LCMV cannot well optimize spatial filters in noisy signals of single trials, but it possibly optimizes well in averaged signals. My question is whether it makes sense to prepare a common spatial filter across conditions using averaged signals, and then apply it to the averaged signal of each condition, with setting cfg.grid.filter? I tried the method using the following script, and its computation finished without any error message. However, I'm afraid that my understanding might be incorrect, i.e., the issue of common filter has nothing to do with the issue of optimization. Could someone please make the issues clearer? 

 

Thank you in advance for your kind responses,

 

 

With best regards,

Masaki Maruyama
 
Inserm U.992 - Neuroimagerie Cognitive 
CEA/SAC/DSV/I2BM/NeuroSpin
Bât 145, Point Courrier 156 
F-91191 GIF/YVETTE, FRANCE 
http://www.unicog.org/

 

 

%Optimize spatial filter in common across conditions and times

for cnd = 1:length(MegFile)

    

    load(MegFile{cnd}); %Load "DataPostICA" containing single trials in a condition

    cfg = [];cfg.removemean = 'no';

    %Optimization of spatial filter based on covariance of averaged field (see Robert email on 04/02/2009)

    cfg.keeptrials = 'no';cfg.covariance = 'no';

    data_temp = timelockanalysis(cfg,DataPostICA);

    if cnd == 1

        data_cmb = data_temp;

    else

        data_cmb = appenddata([],data_cmb,data_temp);

    end

    clear data_temp

end

cfg.keeptrials = 'no';cfg.covariance = 'yes';

cfg.covariancewindow = [-0.5 1];

data = timelockanalysis(cfg,data_cmb);

 

cfg = [];cfg.method = 'lcmv';cfg.grid = grid;cfg.vol = hdm;cfg.lambda = '5%';

cfg.keepfilter = 'yes';

source = sourceanalysis(cfg, data);

common_filter = source.avg.filter;

 

 

%Beamforming using the common spatial filter

for cnd = 1:length(MegFile)

    

    load(MegFile{cnd});%Load "DataPostICA" again

    

    cfg = [];cfg.removemean = 'no';

    cfg.covariancewindow = [-0.5 1];

    cfg.keeptrials = 'no';cfg.covariance = 'no';

    data_temp = timelockanalysis(cfg,DataPostICA);

    cfg.keeptrials = 'no';cfg.covariance = 'yes';

    data = timelockanalysis(cfg,data_temp);

    

    cfg = [];cfg.method = 'lcmv';cfg.grid = grid;cfg.vol = hdm;cfg.lambda = '5%';

    cfg.grid.filter = common_filter;

    source = sourceanalysis(cfg, data);

    save(strcat('SourceAnalysis','_cnd',num2str(cnd)), 'source');

end

 

 


----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20100129/b10619a9/attachment-0001.html>