sourcestatistics: another bug/problem

Robert Oostenveld r.oostenveld at FCDONDERS.RU.NL
Thu Sep 29 09:51:17 CEST 2005


Hi Ho Fei,

You have to specify the experimental design in cfg.design, similar to
the design matrix in SPM. You have multiple source volumes, one for
each subject and condition. For each source volume you have to give
the condition label (as a number), i.e. the length of cfg.design
equals the number of source volumes and the number of unique elements
equals the number of conditions. You can also use more elaborate
design matrices with multiple rows, one row for one each factor, but
it you are not familiar with design matrices, I would not try that
immediately.

best regards,
Robert


On 26-sep-2005, at 16:09, Hoi Fei Kwok wrote:

> Dear Robert,
>
> Thanks. I have changed line 65 of the statistic_wrapper.m but
> encountered
> the following error:
>
> ??? Reference to non-existent field 'design'.
>
> Error in ==> fieldtrip-20050901\private\statistics_wrapper at 98
>   design = cfg.design;
>
> Error in ==> sourcestatistics at 86
>   stat = statistics_wrapper(cfg, varargin{:});
>
> The problem seems to arise from the fact that while calling the
> source_statistics, 'design' is not one of the fields that one has to
> specify.
>
> Regards,
> Hoi Fei
>
>
> -----Original Message-----
> From: FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL]
> On Behalf
> Of Robert Oostenveld
> Sent: 26 September 2005 14:33
> To: FIELDTRIP at NIC.SURFNET.NL
> Subject: Re: [FIELDTRIP] sourcestatistics
>
> Hi Ho Fei
>
> On 16-sep-2005, at 12:08, Hoi Fei Kwok wrote:
>
>> We've been trying to use sourcestatistics.m but I found that first
>> of all a
>> few private functions are missing. These include:
>> sourcestatistics_parametric.m
>> sourcestatistics_randomization.m
>> and sourcestatistics_randcluster.m
>>
>
> There are different methods available for statistical assesements of
> differences between conditions, most of them are implemented for all
> different data types, namely ERPs TFRs and source reconstructions.
> These are implemented as submethods of the sourcestatistics,
> freqstatistics and timelockstatistics functions. These functions will
> call the appropritate subfunction, and these are missing. Some of the
> subfunctions are still work in progress and are not part of the
> release version of fieldtrip (because of scientific novelty).
> However, there are also a few subfunctions that are supposed to be
> part of the release version. I just checked and they are indeed not
> part of the current zip file any more. I will make sure to include
> them in the upcoming daily version. Furthermore, I will also send
> them to you as attachement in a separate mail (probably tomorrow).
>
>
>> Another problem occurred when we tried to set the cfg.approach to
>> 'stat':
>> ...
>> Error in ==> fieldtrip-20050901\private\statistics_wrapper at 65
>>     if isfield(varargin{1}, 'pos') && (length(varargin{i}.pos)
>> ~=length
>> (varargin{1}.pos) || ~all(varargin{i}.pos==varargin{1}.pos))
>>
>
> That is indeed a bug. Please change the last part of the statement
> from "all(varargin{i}.pos==varargin{1}.pos)" into "all(varargin{i}.pos
> (:)==varargin{1}.pos(:))". I will fix it in the official version as
> well.
>
> best regards,
> Robert
>
> PS Sorry for my late reply, I was on the SPR conference.
>
>

=======================================================
Robert Oostenveld, PhD
F.C. Donders Centre for Cognitive Neuroimaging
Radboud University Nijmegen
phone: +31-24-3619695
http://www.ru.nl/fcdonders/



More information about the fieldtrip mailing list