[FieldTrip] DICS beamforming

[Eelke Spaak]

Hi Sara,

Thanks for following up! It's very likely that this is indeed a
version-related issue. cfg.sourcemodel is the correct configuration
parameter, cfg.grid has been deprecated for a long time (several
years) now. So, if you are using a recent version of FieldTrip then it
should indeed be cfg.sourcemodel.

Best,
Eelke



On Thu, 16 Apr 2020 at 20:56, Hussain, Sara (NIH/NINDS) [F]
<sara.hussain at nih.gov> wrote:
>
> I wanted to follow up on this problem in case anyone else was having the same one. I had been following the “Localizing visual gamma and corticomuscular coherence” tutorial, where it stated that the inputs to ft_sourceanalysis should include your leadfield in cfg.sourcemodel. However, after some debugging it become apparent that this was not recognizing the leadfield, and that the leadfield should instead be specific in cfg.grid. This completely solves the problem – and may in fact be a version-related issue.
>
>
>
> Hopefully this can help someone else avoid frustration upon encountering a similar error!
>
>
>
> From: "Hussain, Sara (NIH/NINDS) [F]" <sara.hussain at nih.gov>
> Reply-To: FieldTrip discussion list <fieldtrip at science.ru.nl>
> Date: Friday, April 10, 2020 at 2:28 PM
> To: "fieldtrip at science.ru.nl" <fieldtrip at science.ru.nl>
> Subject: [FieldTrip] DICS beamforming
>
>
>
> Hi all –
>
>
>
> I am trying to source localize some broadband oscillatory data using the ‘dics’ beamforming option, as documented in the tutorial “Localizing visual gamma and corticomuscular coherence.” I am currently using standard mri and segmentation via the fieldtrip templates, and generating a dipole grid with 500 dipoles using the ‘basedonvol’ approach. I then compute the leadfield based on the positions of these dipoles. Up to this point, everything seems ok and checks out during plotting. But when I try to run ft_sourceanalysis I am getting the following error:
>
>
>
> Warning: dipole lies on boundary of volume model
>
> > In inf_medium_leadfield (line 54)
>
>   In eeg_leadfieldb (line 92)
>
>   In ft_compute_leadfield (line 443)
>
>   In beamformer_dics (line 305)
>
>   In ft_sourceanalysis (line 674)
>
> Error using svd
>
> Input to SVD must not contain NaN or Inf.
>
>
>
> Error in beamformer_dics>pinv (line 643)
>
>   [U,S,V] = svd(A,0);
>
>
>
> Error in beamformer_dics (line 336)
>
>         filt = pinv(lf' * invCf * lf) * lf' * invCf;              % Gross eqn. 3, use PINV/SVD
>
>         to cover rank deficient leadfield
>
>
>
> Error in ft_sourceanalysis (line 674)
>
>           dip(i) = beamformer_dics(grid, sens, headmodel, [],  squeeze_Cf, optarg{:});
>
>
>
>
>
> I’ve seen this error asked in other threads but I’ve tried what those suggest with no luck. I’m assuming this has something to do with the fact that the dipoles aren’t fully embedded in the grey matter – but I’ve tried changing the inward shift parameter etc during grid generation and it hasn’t solved it. Any suggestions?  Code is below..
>
>
>
> ===================
>
>
>
> % housekeeping
>
> clearvars
>
> basedir='/net/nindsdirfs.ninds.nih.gov/ifs/shares/HCPS/';
>
> addpath([basedir 'SaraHussain/Analysis_Tools/fieldtrip-20160718']);
>
> ft_defaults;
>
>
>
> % read standard mri
>
> cd([basedir 'SaraHussain/Analysis_Tools/fieldtrip-20160718/template/headmodel']);
>
> mri=load('standard_mri.mat');
>
> mri=mri.mri;
>
>
>
> % read standard segmentation
>
> seg=load('standard_seg.mat');
>
> seg=seg.mri;
>
>
>
> % read standard headmodel
>
> vol=load('standard_bem.mat'); %vol
>
> vol=vol.vol;
>
>
>
> % create sourcemodel/grid based on standard mri
>
> cfg=[];
>
> cfg.headmodel=vol;
>
> cfg.moveinward=2;
>
> cfg.grid.unit='mm';
>
> grid=ft_prepare_sourcemodel(cfg);
>
>
>
> % check segmentation
>
> seg.transform=mri.transform;
>
> seg.anatomy=mri.anatomy;
>
> cfg=[];
>
> cfg.funparameter='gray';
>
> ft_sourceplot(cfg,seg);
>
> pause; close all;
>
>
>
> % check volume conduction model
>
> figure; ft_plot_mesh(vol.bnd(3),'facecolor','none'); %brain
>
> figure; ft_plot_mesh(vol.bnd(2),'facecolor','none'); %skull
>
> figure; ft_plot_mesh(vol.bnd(1),'facecolor','none'); %scalp
>
> pause; close all;
>
>
>
> % check electrode positions
>
> figure;
>
> elec=ft_read_sens('standard_1005.elc','senstype','eeg');
>
> ft_plot_mesh(vol.bnd(3),'edgecolor','none','facecolor',[0.2 0.2 0.2],'facealpha',0.4,'edgecolor','none','edgealpha',0.05); hold on;
>
> ft_plot_mesh(vol.bnd(2),'edgecolor','none','facealpha',0.4);
>
> ft_plot_mesh(vol.bnd(1),'edgecolor','none','facecolor',[0.4 0.6 0.4]);
>
> ft_plot_sens(elec,'label','yes');
>
> pause; close all
>
>
>
> % check grid position
>
> figure;
>
> ft_plot_mesh(vol.bnd(3),'edgecolor','none'); hold on; alpha 0.4; %brain
>
> ft_plot_mesh(grid.pos(grid.inside,:));
>
> pause; close all;
>
>
>
> % load data
>
> cd([basedir 'SaraHussain/SS_42H_63/clean_tep_data']);
>
> load([peak_subj{jjj} '_double_cleaned_peak.mat']);
>
>
>
> % demean/detrend and update channel names
>
> cfg=[];
>
> cfg.demean='yes';
>
> cfg.detrend='yes';
>
> data=ft_preprocessing(cfg,clean_data);
>
> data.label=labels;
>
>
>
> % select relevant post-stim latencies and channels
>
> cfg=[];
>
> short=ft_selectdata(cfg,data);
>
>
>
> % calculate post-stim frequency using mtmfft approach
>
> cfg=[];
>
> cfg.method='mtmfft';
>
> cfg.output='powandcsd';
>
> cfg.tapsmofrq=48;
>
> cfg.foi=52;
>
> peak=ft_freqanalysis(cfg,short);
>
>
>
> % create leadfield
>
> cfg=[];
>
> cfg.headmodel=vol;
>
> cfg.grid.pos=grid.pos;
>
> cfg.grid.inside=grid.inside;
>
> cfg.elec=elec;
>
> lead=ft_prepare_leadfield(cfg,peak);
>
>
>
> % get source localized data
>
> cfg=[];
>
> cfg.elec=elec;
>
> cfg.method='dics';
>
> cfg.sourcemodel=lead;
>
> cfg.headmodel=vol;
>
> source_peak=ft_sourceanalysis(cfg,peak);
>
>
>
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202