[Eeglablist] Removing Components makes "index exceeds matrix dimension" error in SIFT

[boglyas]

Thank you very much,  I'll try these methods tomorrow.

Balázs
2013.08.07. 20:01, "Tim Mullen" <mullen.tim at gmail.com> ezt írta:

> Hi Balazs,
>
> First, just to be clear, you pruned the set of components using
> pop_subcomp (which will ensure a consistent dataset with proper dimensions
> for .icawinv and equal number of dipoles to components)?
>
> If so, I suspect you ran SIFT once before for this dataset before pruning.
> In this case the pre-processing function would have retrieved the default
> value for ComponentsToKeep from the currently stored value in
> EEG.CAT.curComponentNames -- i.e. the last set of components you selected
> in SIFT before you pruned the dataset. This would explain why the default
> list of components is set to 1:16. BrainMovie3D uses the list in
> EEG.CAT.curComponentNames to plot labels at each node location. If the
> length of this list exceeds the actual number of nodes (components) an
> error will occur.
>
> However, the fix is simple.
>
> The cleanest, failsafe, solution is just to remove the 'CAT' field from
> the EEG structure...
> EEG = rmfield(EEG,'CAT');
> ALLEEG(CURRENTSET) = EEG;
> ...and then run the SIFT pipeline again. This will clear all SIFT usage
> history (note this will also clear any previous SIFT result you had).
>
> If you want to preserve your previously calculated results, then you
> *should* only need to fix the .curComponentNames, and .curComps lists by
> typing:
> EEG.CAT.curComponentNames = strtrim(cellstr(num2str((1:9)'))');
> EEG.CAT.curComps = 1:9;
> ALLEEG(CURRENTSET) = EEG:
>
> The BrainMovie should now work properly.
>
> A tip: in many cases there is automatic "sanity checking" of default
> options, but in some cases these checks may not be present. So always make
> sure that default options -- such as the range of components -- match your
> expectations before proceeding to the next step in processing. In releases
> of SIFT prior to 1.0b, it's also important not to run pre-processing twice
> on the same dataset.
>
> Cheers,
> Tim
>
>
> On Tue, Aug 6, 2013 at 9:20 AM, boglyas <kamanbalazs at gmail.com> wrote:
>
>> Hy there,
>>
>> I tried to remove components from EEG data set for better results, but I
>> cannot use the 3D brain movie in SIFT after that,because it gives an error
>> "Index exceeds matrix dimensions".
>>
>> Basically it was a 61 channel recording with 60 (or 61) components, but I
>> had a data set that was already pruned, it had only 16 components/dipoles.
>> Some of the dipoles from the 16 were not in the cortex, they were deep
>> inside the brain (eg. thalamus), so I decided to remove those in order to
>> have better results later. After removing these components, if i check the
>> plotting in the EEGlab, or the  EEG <1x1 struct> in MATLAB, it all says
>> that it has 9 components left.
>>
>> At the same time, if I try to run SIFT, every tool thinks that i have 16
>> components (and not 9). Even in pre-processing, the default "components to
>> keep" goes like: 1; 2; 3; 4; 5; 6; 7; 8; 9; 10; 11; 12; 13; 14; 15; 16;
>>
>> Meanwhile it doesn't give error during the pre-pocessing or the model
>> fitting, and the validation also seems ok. I can visualize the connectivity
>> result in a Time-Frequency grid (and it only plots 9 x 9 figure), but when
>> I try to visualize in BrainMovie3D, it crashes, error messages, and so on.
>>
>> What should I do, or what did I do wrong?
>>
>> Thanks,
>> Balazs
>>
>>
>> 2013/8/6 boglyas <kamanbalazs at gmail.com>
>>
>>> Hy there,
>>>
>>> I tried to remove components from EEG data set for better results, but I
>>> cannot use the 3D brain movie in SIFT after that,because it gives an error
>>> "Index exceeds matrix dimensions".
>>>
>>> Basically it was a 61 channel recording with 60 (or 61) components, but
>>> I had a data set that was already pruned, it had only 16
>>> components/dipoles. Some of the dipoles from the 16 were not in the cortex,
>>> they were deep inside the brain (eg. thalamus), so I decided to remove
>>> those in order to have better results later. After removing these
>>> components, if i check the plotting in the EEGlab, or the  EEG <1x1 struct>
>>> in MATLAB, it all says that it has 9 components left.
>>>
>>> At the same time, if I try to run SIFT, every tool thinks that i have 16
>>> components (and not 9). Even in pre-processing, the default "components to
>>> keep" goes like: 1; 2; 3; 4; 5; 6; 7; 8; 9; 10; 11; 12; 13; 14; 15; 16;
>>>
>>> Meanwhile it doesn't give error during the pre-pocessing or the model
>>> fitting, and the validation also seems ok. I can visualize the connectivity
>>> result in a Time-Frequency grid (and it only plots 9 x 9 figure), but when
>>> I try to visualize in BrainMovie3D, it crashes, error messages, and so on.
>>>
>>> What should I do, or what did I do wrong?
>>>
>>> Thanks,
>>> Balazs
>>>
>>
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to
>> eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to
>> eeglablist-request at sccn.ucsd.edu
>>
>
>
>
> --
> ---------  αντίληψη -----------
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130807/a5bf1fcd/attachment.html>