[FieldTrip] ft_sourcestatistics startup memory problem

[Johanna Zumer]

Hi Akiko,

The .pos entries are (as expected) different for every subject due to
the different coregistration of that subject to the standard MNI.
However, the grid positions all originate from the same original
MNI-based positions, and are in the same order.  Thus,
d1group{1}.pos(1,:) corresponds to the template_grid.pos(1,:) and so
forth.   You can substitue d1group{n}.pos=template_grid.pos for all
subjects, and then call the stats function, and the results are now in
the MNI template grid space.

Cheers,
Johanna

2012/10/22 Akiko Ikkai <akiko.ikkai at gmail.com>:
> Hi Johanna,
>
> you are right! specifying cfg.grid = grid; was the key; ft_sourceanalysis
> for each condition runs to the completion :)
>
> However, since my data is EEG and the electrode locations vary (hence the
> grid coordinates vary) between subjects, running stats across subjects
> without normalizing might be causing a problem. For example, the first 3
> points in subject1's .pos are (after creating template grid etc.., and run
> ft_sourceanalysis using MNI grid):
>
> d1group{1}.pos(1:3,:)
> ans =
>    -8.7000   -7.5000   -7.6000
>    -7.7000   -7.5000   -7.6000
>    -6.7000   -7.5000   -7.6000
>
> and the same points for the subject2's are:
>
> d1group{2}.pos(1:3,:)
> ans =
>    -8.3000   -7.6000   -8.2000
>    -7.3000   -7.6000   -8.2000
>    -6.3000   -7.6000   -8.2000
>
> So when I try to run ft_sourcestatistics. I get an error message:
> ??? Error using ==> statistics_wrapper at 109
> grid locations of the source reconstructions do not match, use
> NORMALISEVOLUME
> first
>
> Error in ==> ft_sourcestatistics at 100
>     [stat, cfg] = statistics_wrapper(cfg, varargin{:});
>
> Perhaps I should interpolate and normalize each subject, and run group stats
> afterall?
>
> Thanks! Akiko
>