[FieldTrip] ft_sourceanalysis using LCMV beamformer and grid from ft_prepare_leadfield

[Tyler Grummett]

Fieldtrip,


Ive got it working now.


The follow error:


*****************************************************************


Error using  *
Inner matrix dimensions must agree.

Error in beamformer_lcmv (line 268)
    filt = pinv(lf' * invCy * lf) * lf' * invCy;              % van Veen eqn. 23, use PINV/SVD to
    cover rank deficient leadfield

Error in ft_sourceanalysis (line 794)
      dip(i) = beamformer_lcmv(grid, sens, vol, squeeze_avg, squeeze(Cy(i,:,:)), optarg{:});

*****************************************************************


It is because:

if is 128x3

invCy is 124x124


When I ran the code:


            cfg = [];
            cfg.channel = 'EEG'; % only works for preprocessing
            cfg.vartrllength = 2;  % accept variable length trials
            cfg.covariance = 'yes';
            cfg.covariancewindow = 'all';
            cfg.keeptrials = 'no';
            timelock = ft_timelockanalysis(cfg, data);


It did the timelockanalysis on 124 channels, but it didnt remove the non-EEG channels from timelock.elec

and so the four EOG channels explains the 128.


timelock =

       avg: [124x11758 double]
       var: [124x11758 double]
      time: [1x11758 double]
       dof: [124x11758 double]
     label: {1x124 cell}
    dimord: 'chan_time'
       cov: [124x124 double]
      elec: [1x1 struct]
       cfg: [1x1 struct]

timelock.elec =

    chanpos: [128x3 double]
    elecpos: [128x3 double]
      label: {1x128 cell}
       type: 'eeg1005'
       unit: 'mm'

I dont know if this is a bug or not. Please advise.

Tyler




*************************

Tyler Grummett ( BBSc, BSc(Hons I))
PhD Candidate
Brain Signals Laboratory
Flinders University
Rm 5A301
Ext 66124
________________________________
From: fieldtrip-bounces at science.ru.nl <fieldtrip-bounces at science.ru.nl> on behalf of Tyler Grummett <tyler.grummett at flinders.edu.au>
Sent: Tuesday, 8 April 2014 12:18 PM
To: FieldTrip discussion list
Subject: [FieldTrip] ft_sourceanalysis using LCMV beamformer and grid from ft_prepare_leadfield


Hello fieldtrip,


I am trying to perform an LCMV beamformer on timelockanalysis data.


timelock =

       avg: [124x20517 double]
       var: [124x20517 double]
      time: [1x20517 double]
       dof: [124x20517 double]
     label: {1x124 cell}
    dimord: 'chan_time'
       cov: [124x124 double]
      elec: [1x1 struct]
       cfg: [1x1 struct]

I first run ft_prepare_leadfield

            cfg                 = [];
            cfg.elec            = timelock.elec;
            cfg.vol             = vol;
            cfg.grid.resolution = 1;   % use a 3-D grid with a 1 cm resolution
            cfg.grid.unit       = 'cm';
            cfg.normalize       = 'yes'; % if you are not contrasting the activity of interest again another condition or baseline time-window
            grid = ft_prepare_leadfield( cfg);

grid =
        xgrid: [-5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8]
        ygrid: [-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7]
        zgrid: [0 1 2 3 4 5 6 7 8 9 10 11 12]
          dim: [14 14 13]
          pos: [2548x3 double]
         unit: 'cm'
       inside: [1x1409 double]
      outside: [1x1139 double]
          cfg: [1x1 struct]
    leadfield: {1x2548 cell}

I then try to run the ft_sourceanalysis code:

            cfg              = [];
            cfg.channel = timelock.label;
            cfg.method       = 'lcmv';
            cfg.grid = grid;
            cfg.vol          = vol;
            cfg.keepfilter = 'yes';
            source = ft_sourceanalysis( cfg, timelock);


And it crashes with the error message:

Error using  *
Inner matrix dimensions must agree.

Error in beamformer_lcmv (line 268)
    filt = pinv(lf' * invCy * lf) * lf' * invCy;              % van Veen eqn. 23, use PINV/SVD to
    cover rank deficient leadfield

Error in ft_sourceanalysis (line 794)
      dip(i) = beamformer_lcmv(grid, sens, vol, squeeze_avg, squeeze(Cy(i,:,:)), optarg{:});



However, the code does work if it is:

            cfg              = [];
            cfg.channel = timelock.label;
            cfg.method       = 'lcmv';
            cfg.grid.inside = grid.inside;
            cfg.vol          = vol;
            cfg.keepfilter = 'yes';
            source = ft_sourceanalysis( cfg, timelock);

However, I run the sourceplot code produces something that looks like the picture attached.

What is interesting is that when I run the interp code:

    the input is source data with 16777216 positions on a [256 256 256] grid
    not plotting anatomy

But when you run the same code from sourcePost_nocon from the beamformer tutorial,  you get the following when you run the code:

    the input is source data with 3042 positions on a [18 13 13] grid
    the input is volume data with dimensions [256 256 256]

I have used the same mri file for both.
?
Please help, I cant work out the issue.

Tyler



*************************

Tyler Grummett ( BBSc, BSc(Hons I))
PhD Candidate
Brain Signals Laboratory
Flinders University
Rm 5A301
Ext 66124
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140408/fb78308e/attachment-0001.html>