[FieldTrip] source analysis- NaN in leadfield matrices

[Johanna Zumer]

Hi Doriana,

It looks like you're using a FieldTrip verison from 20140424.   There was a
problem last year with bemcp causing NaNs, but that has been fixed.  Could
you try updating to a recent version and see if that solves it?

Cheers,
Johanna
On 22 Mar 2015 17:22, "Doriana De Marco" <doriana.demarco at gmail.com> wrote:

> Dear Fieldtrip members,
>
> I’m dealing with source analysis and I have a big problem that I can’t
> resolve…
>
> Filedtrip gives me an error after launching of ft_sourceanalysis:
>
> ??? Error using ==> svd
>
> Input to SVD must not contain NaN or Inf.
>
> Error in ==> rank at 15
>
> s = svd(A);
>
> Error in ==> ft_eloreta at 132
>
>     rank_lf(i) = rank(dip.leadfield{i});
>
> Error in ==> ft_sourceanalysis at 850
>
>       dip(i) = ft_eloreta(grid, sens, vol, squeeze(avg(i,:,:)),
> squeeze(Cy(i,:,:)),
>
>       optarg{:});
>
> In my leadfield computation there are in fact many NaN points.  I report
> my script below because I can’t find where the error can be.  I used
> standard_mri template to prepare head model.
>
> %% 1. PREPARING HEAD MODEL
>
> %METHOD 1
>
> %if you have a structural mri and you want to create headmodel
>
>
>
> %load headmodel – mri
>
> load
> ('C:\Users\DorianaNote\Documents\MATLAB\fieldtrip-20140424\fieldtrip-20140424\template\headmodel\standard_mri');
>
> cfg          = [];
>
> cfg.output    = {'brain','skull','scalp'};
>
> template_seg = ft_volumesegment(cfg, mri);
>
>
>
> cfg          = [];
>
> cfg.method   = 'bemcp';
>
> template_vol = ft_prepare_headmodel(cfg, template_seg);
>
>
>
> elec = ft_read_sens('GSN-HydroCel-128.sfp');
>
> template_vol = ft_convert_units(template_vol, elec.unit);
>
>
>
> cfg           = [];
>
> cfg.method    = 'interactive';
>
> cfg.elec      = elec;
>
> cfg.headshape = template_vol.bnd(3);
>
> elec_aligned  = ft_electroderealign(cfg);
>
>
>
> cfg = [];
>
> cfg.grid.unit = template_vol.unit;
>
> cfg.elec = elec_aligned;
>
> cfg.mri = mri;
>
> cfg.vol = template_vol;
>
> grid = ft_prepare_sourcemodel(cfg)
>
>
>
> cfg                 = [];
>
> cfg.elec            = elec_aligned;
>
> cfg.vol             = template_vol;
>
> cfg.reducerank      = 3;
>
> cfg.grid            = grid;
>
> cfg.channel         = {'all'};
>
> cfg.grid.unit       = 'cm';
>
> [grid] = ft_prepare_leadfield(cfg, data);
>
>
>
> cfg              = [];
>
> cfg.method       = 'eloreta';
>
> cfg.grid         = grid;
>
> cfg.elec         = elec_aligned;
>
> cfg.vol          = template_vol;
>
> source = ft_sourceanalysis(cfg, data);
>
>
>
> I’m trying to analyze component data in order to localize comp.topo
> parameter. I used the function comp2timelocked to transform data in a
> timelock-like structure... I don’t know if it can be a problem.
>
>
> I sincerely hope someone of you can help me
>
> Many thanks
>
>  ----------------
>
> Doriana De Marco, Ph.D. Student
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150323/7f6cd75a/attachment.html>