[FieldTrip] problem with clusterplot?

"Jörn M. Horschig" jm.horschig at donders.ru.nl
Mon Apr 2 16:34:16 CEST 2012


Dear Mimma,

okay, I see. First of all, behaviour of ft_clusterplot is perfectly 
fine. The way ft_freqstatistics works is that it first computes a 
t-value for all datapoints, and then computes clusters in time, space 
and/or frequency, depending on whether you set cfg.avgovertime or 
cfg.avgoverfreq to 'yes'. All these t-values are uncorrected t-values, 
therefore the reliability of the statistics will suffer from multiple 
comparisons (as a side note, this might be an interesting read: 
http://www.danielbor.com/dilemma-weak-neuroimaging/ ).
The test you specified will compute permutations of your data and then 
compare the actual observed t-value (as described above) with the 
distribution of the random permutations. Only if your observation lies 
in the significant head or tail of the permutation distribution (as 
defined by the first/last 2.5%, i.e. your alpha value), the found 
cluster will be considered to be significant. Thus, even if the t-value 
might be significant, it will be corrected thereafter by the cluster 
permutation. For a detailled description, I would advise you to read 
this paper on which the whole method is based on:
http://www.sciencedirect.com/science/article/pii/S016502700700074X

In your data, the observed cluster is not significant anymore after the 
cluster-based correction. You can also check it by typing 
stat.posclusters.prob, which will return a value > 0.025

Hope I described it somewhat comprehensible.
Best,
Jörn


On 4/2/2012 11:11 AM, veniero at med.unibs.it wrote:
> Dear Jörn,
> I'm copmparing two conditions gavgM1_post5 and gavgM1_pre5.
> Here the settings I'm using:
>
> cfg = [];
> cfg.channel          = {'EEG'};
> cfg.layout= 'biosemi64.lay';
> cfg.latency          = 'all';
> cfg.frequency        = [20 20];
> cfg.avgeoverfreq     = 'yes';
> cfg.method           = 'montecarlo';
> cfg.statistic        = 'indepsamplesT';
> cfg.correctm         = 'cluster';
> cfg.clusteralpha     = 0.05;
> cfg.clusterstatistic = 'maxsum';
> cfg.minnbchan        = 2;
> cfg.tail             = 0;
> cfg.clustertail      = 0;
> cfg.alpha            = 0.025;
> cfg.numrandomization = 500;
> design = zeros(1,size(gavgM1_post5.powspctrm,1) +
> size(gavgM1_pre5.powspctrm,1));
> design(1,1:size(gavgM1_post5.powspctrm,1)) = 1;
> design(1,(size(gavgM1_post5.powspctrm,1)+1):(size(gavgM1_post5.powspctrm,1)+...
>    size(gavgM1_pre5.powspctrm,1))) = 2;
> cfg.design           = design;
> cfg.ivar             = 1;
>
> [stat] = ft_freqstatistics(cfg, gavgM1_post5, gavgM1_pre5);
>
>
> For the clusterplot:
> stat.raweffect = gavgM1_post5.powspctrm - gavgM1_pre5.powspctrm;
>
> cfg = [];
> cfg.alpha  = 0.025;
> cfg.zparam = 'raweffect';
> cfg.zlim   = [-1 1];
> cfg.layout = 'biosemi64.lay';
> ft_clusterplot(cfg, stat);
>
> Many thanks,
> Mimma
>
>
>
>> Dear Mimma,
>>
>> Most likely, you specified a different alpha when calling
>> ft_freqstatistics than when calling ft_clusterplot. Could you paste the
>> cfg settings for both functions here, so that we can try to track down
>> where exactly this error comes from?
>>
>> Best,
>> Jörn
>>
>>
>> On 4/1/2012 7:03 PM, veniero at med.unibs.it wrote:
>>> Dear FieldTrip  users,
>>> I'm trying to run a cluster-based permutation tests on time-frequency
>>> data. I get the result:
>>> found 1 positive clusters in observed data found 0 negative clusters in
>>> observed data, but I'm not able to use clusterplot because if the
>>> following error:
>>> ??? Error using ==>   ft_clusterplot at 171
>>> no clusters present with a p-value lower than the specified alpha,
>>> nothing
>>> to plot
>>>
>>> Is this possible? Maybe it's just a trivial question.
>>>
>>> Many thanks,
>>> Mimma
>>>
>>>
>>> _______________________________________________
>>> fieldtrip mailing list
>>> fieldtrip at donders.ru.nl
>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>
>> --
>> Jörn M. Horschig
>> PhD Student
>> Donders Institute for Brain, Cognition and Behaviour
>> Centre for Cognitive Neuroimaging
>> Radboud University Nijmegen
>> Neuronal Oscillations Group
>>
>> P.O. Box 9101
>> NL-6500 HB Nijmegen
>> The Netherlands
>>
>> Contact:
>> E-Mail: jm.horschig at donders.ru.nl
>> Tel:    +31-(0)24-36-68493
>> Web: http://www.ru.nl/donders
>>
>> Visiting address:
>> Trigon, room 2.30
>> Kapittelweg 29
>> NL-6525 EN Nijmegen
>> The Netherlands
>>
>> _______________________________________________
>> fieldtrip mailing list
>> fieldtrip at donders.ru.nl
>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip


-- 
Jörn M. Horschig
PhD Student
Donders Institute for Brain, Cognition and Behaviour
Centre for Cognitive Neuroimaging
Radboud University Nijmegen
Neuronal Oscillations Group

P.O. Box 9101
NL-6500 HB Nijmegen
The Netherlands

Contact:
E-Mail: jm.horschig at donders.ru.nl
Tel:    +31-(0)24-36-68493
Web: http://www.ru.nl/donders

Visiting address:
Trigon, room 2.30
Kapittelweg 29
NL-6525 EN Nijmegen
The Netherlands




More information about the fieldtrip mailing list