[FieldTrip] Can the warning "Warning: could not determine dimord of "comp" in:" influence the results?

[Eelke Spaak]

Dear Helena,

Yes, you can safely ignore the warning.

Best,
Eelke

On Tue, 9 Jun 2020 at 12:14, Helena Pereira
<hri.pereira at campus.fct.unl.pt> wrote:
>
> Dear Eelk,
>
> Perhaps everything is ok, and I can ignore the warning. The structure of my data is as follows (and contains the components that you mentioned: e.g. .topolabel), or it should also appear in the component analysis?
>
> ica_cleaned =
>   struct with fields:
>        fsample: 512
>           time: {1×100 cell}
>          trial: {1×100 cell}
>           topo: [69×69 double]
>       unmixing: [69×69 double]
>          label: {69×1 cell}
>      topolabel: {69×1 cell}
>     sampleinfo: [100×2 double]
>      trialinfo: [100×1 double]
>            cfg: [1×1 struct]
>
> Thank you for your time and help.
>
> Best,
> Helena Pereira
>
>
> Helena Pereira <hri.pereira at campus.fct.unl.pt> escreveu no dia terça, 9/06/2020 à(s) 09:38:
>>
>> Dear Eelk,
>>
>> Yes, 'ica_cleaned' is produced directly from ft_componentanalysis:
>>
>> cfg = [];
>> cfg.dataset=subject;
>> cfg.trialfun = 'ft_trialfun_general'; % this is the default
>> cfg.trialdef.eventtype  = 'Stimulus';
>> cfg.trialdef.eventvalue ={Stimulus}';
>> cfg.trialdef.prestim    = 2; %latency in seconds
>> cfg.trialdef.poststim   = 2; %latency in seconds
>> cfg=ft_definetrial(cfg);
>>
>> cfg.channel = [1:69]; %discard channels not used
>> cfg.preproc.reref = 'yes';
>> cfg.preproc.refchannel = {'M1' 'M2'}; %avg mastoids
>> cfg.continuous = 'yes';
>> cfg.preproc.detrend = 'no';
>> %Baseline correction
>> %cfg.preproc.demean = 'yes';
>> cfg.prepoc.baselinewindow  = [-0.2 0];
>> %Filtering
>> cfg.preproc.lpfilter = 'yes';
>> cfg.preproc.hpfilter = 'yes';
>> cfg.preproc.lpfreq = 100;
>> cfg.preproc.hpfreq = 1;
>> data=ft_preprocessing(cfg);
>>
>> cfg = [];
>> cfg.method = 'runica';
>> cfg.demean = 'yes';
>> ica_cleaned = ft_componentanalysis(cfg,data);
>>
>> % %Save ICA data
>> % outfile=strcat(save_dir,'/',infile,'_', sti,'.mat')
>> % save(outfile, 'ica_cleaned', '-v7.3')
>>
>> cfg = [];
>> cfg.component = 1:20;
>>
>> cfg.layout = 'biosemi64.lay';
>> cfg.comment   = 'no';
>> figure();
>> ft_topoplotIC(cfg,ica_cleaned)
>>
>> The data is in the BrainVision Format ( .vhdr, .vmrk and .seg), obtained as suggested by Fieldtrip website: http://www.fieldtriptoolbox.org/getting_started/brainvision/
>>
>> Can I ignore the Warning?
>>
>> Best,
>> Helena Pereira
>>
>> Eelke Spaak <e.spaak at donders.ru.nl> escreveu no dia terça, 9/06/2020 à(s) 08:15:
>>>
>>> Dear Helena,
>>>
>>> In general, you can probably safely ignore the warning. It just means
>>> that FieldTrip is unsure of the dimensionality of the 'comp' field,
>>> because there is no 'compdimord' and also the dimensionality does not
>>> match the data in any straightforward way.
>>>
>>> However, the structure of your componentanalysis output is somewhat
>>> strange. From .topo and .unmixing, it appears there is only one
>>> component. However, .label has 69 elements, and there is no .topolabel
>>> (which usually is there). Is the output you posted directly produced
>>> by ft_componentanalysis (i.e. it is the 'ica_cleaned' variable
>>> immediately after the step you showed)?
>>>
>>> Best,
>>> Eelke
>>>
>>> On Mon, 8 Jun 2020 at 21:42, Helena Pereira
>>> <hri.pereira at campus.fct.unl.pt> wrote:
>>> >
>>> > Dear FieldTrip users,
>>> >
>>> > I'm running the ft_componentanalysis function after applying ICA to my data. Although I'm not getting any error, the following warning came out:
>>> >
>>> > Warning: could not determine dimord of "comp" in:
>>> >
>>> >              label: {69×1 cell}
>>> >                cfg: [1×1 struct]
>>> >               topo: [69×1 double]
>>> >         topodimord: 'chan_comp'
>>> >               comp: 7
>>> >           unmixing: [1×69 double]
>>> >     unmixingdimord: 'chan_topochan'
>>> >
>>> > Can this influence the way the components are displayed and consequently my results? Is this associated with the layout that I used or with the "demean" option?
>>> >
>>> > My code is as follows:
>>> > cfg = [];
>>> > %cfg.channel = 1:64;
>>> > cfg.method = 'runica';
>>> > cfg.demean = 'yes';
>>> > ica_cleaned = ft_componentanalysis(cfg, data);
>>> >
>>> > cfg = [];
>>> > cfg.component = 1:20;
>>> > cfg.layout = 'biosemi64.lay';
>>> > cfg.comment   = 'no';
>>> > ft_topoplotIC(cfg, ica_cleaned )
>>> >
>>> > Thanks in advance,
>>> > Keep safe,
>>> > Helena Pereira
>>> >
>>> >
>>> >
>>> >
>>> >
>>> >
>>> > _______________________________________________
>>> > fieldtrip mailing list
>>> > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>> > https://doi.org/10.1371/journal.pcbi.1002202
>>>
>>> _______________________________________________
>>> fieldtrip mailing list
>>> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>> https://doi.org/10.1371/journal.pcbi.1002202
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202