Problems with second level source statistics for multi plew subjects and firsty level stats - solved :-)

[Jess R. Kerlin]

Dear FTrippers,

I've been replicating the beamformer tutorial to the letter, and arrived
at problem which may be related to the one described below.

Running :
stat = sourcestatistics(cfg, sourceSTAT);

Gives the error:

???  Index of element to remove exceeds matrix dimensions.

Error in ==>
/home/jrkerlin/matlab/EEGLAB/fieldtrip-0.9.8/private/statistics_wrapper.m
(get_source_trial)
On line 443  ==>     dat((j-1).*dim(1) + varargin{1}.outside, :) = [];

Error in ==>
/home/jrkerlin/matlab/EEGLAB/fieldtrip-0.9.8/private/statistics_wrapper.m
On line 179  ==>     [dat, cfg] = get_source_trial(cfg, varargin{:});

Error in ==> /home/jrkerlin/matlab/EEGLAB/fieldtrip-0.9.8/sourcestatistics.m
On line 127  ==>   stat = statistics_wrapper(cfg, varargin{:});

My sourceSTAT.trial(1).pow is a 1 by 2652 double array (so the dim in
get_source_trial is the same) , which may or may not be the proper
dimensions for pow (sourceSTAT.dim = [17 13 12] ). If not, exactly
how/where would I add the proper dim field?

I've also tried running  freqanalysis and sourcestatistics  with a
recent release (20081218), with the same results.

Any advice would be greatly appreciated.

Thanks,
Jess

Michael Wibral wrote:
> Dear listusers,
>
> in several previous mails I have reported about problems occuring when:
>
> 1) doing single subjects source analysis using reverse-normalized grids (thanks to Ingrid's tutorial), common filters for the two conditions and generating single trial source data
> 2) then performing single subject sourcestatistics (getting a sparse data representation with a stat field)
> 3) then doing second level sourcestatistics (which didn't work)
>
> I am happy to report these problems are solved :-). I actually just overlooked a WARNING:... when doing single subject sourceanalysis and sourcestatistics (1st level) is says that you yourself have to add a dim field by hand, when trying to use clusterrandanalysis/sourcestatistics later. I now simply added a dim field to both, to the data and to the cfg for the second level source statistics. The correct choice for this dim field are the dimension of the original, orthogonal, regular grid in MNI space and everything works now.
>
>  I just didn't realize that dim can't be all that easily generated automatically by fieldtrip for sparse data and *warped* grids. This is because the warping could - in principle - be such that dim doesn't make any sense anymore, e.g. because points (almost) switch places along a certain dimension (if they differ in the other coordinates). So neighborhood relations are not well defined a priori in sparse data. Hence, the need for a manual addition of dim in this case (please correct me me if any of the above is wrong).
>
> The version I used and that I can confirm to work was 20081208.
>
> Many thanks for all your support,
> Michael
>
>
>
> ----------------------------------
> The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.
>

----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.