[FieldTrip] Plotting parametric cluster permutation

[Schoffelen, J.M. (Jan Mathijs)]

Hi Monalisa,

The cfg that you used to instruct the statistics computation is somewhat ambiguous, and contains specification for options that are mutually exclusive.
Specifically, when specifying cfg.correctm = ‘fdr’, all options that pertain to clustering have become non-functional (because the thing you fill in for correctm takes precedence in determining what’s going to happen). In other words, you either use ‘fdr’(=false discovery rate), or ‘cluster’(=for cluster-based inference + multiple comparison correction), not both. Thus, indeed, the output of the statistics computation by design will not have any information pertaining to clusters. The effect of the false discovery rate will be visible in the ‘mask’ field, which specifies for each channel/source-time-(frequency) point whether the null hypothesis of exchangeability between conditions can be rejected with sufficient confidence. The relationship between the p-values and mask are not so straightforward when using ‘fdr’ as a correction method. This is something that recently came up in another (unresolved, probably TLDR) thread: https://mailman.science.ru.nl/pipermail/fieldtrip/2024-March/042417.html<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fpipermail%2Ffieldtrip%2F2024-March%2F042417.html&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C9c4ab2b947b347f889a008dc546dd5da%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638478077945050349%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=fHqrXqfBaYWkJoGyr17Z496%2BuyCDk8gWHh31bqNDll4%3D&reserved=0>

So, long story short, if you want to make a visualization of the effect, I would display the statistics’ value (as present in the stat-field) and creatively think of a way of highlighting the data points that survive the statistical mask.

One final point: when using ‘fdr’ typically it only makes sense if you use a very large number of randomizations, in your case I would go way beyond the listed 500. If your design admits (i.e. the maximum number of randomizations will be 2^#-of-participants) you should use at least 10000 or so. If that’s not possible, probably you’d need to consider another way of doing your inference.

Good luck,
Jan-Mathijs


On 4 Apr 2024, at 01:08, MONALISA CHIKEZIE via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:

Hi All,
I am relatively new to Fieldtrip and I want to plot source statistics results. How do I plot the results of my parametric cluster permutation.
I used the following code to run the source statistics.

cfg.stat = [];
cfg.stat.method = 'analytic'; %
cfg.stat.statistic = 'ft_statfun_depsamplesT';
cfg.stat.correctm = 'fdr'; % Analytical methods with an FDR correction
cfg.stat.parameter = 'pow';
cfg.stat.clusteralpha = 0.05;
cfg.stat.clustercritval = 0.05;
cfg.stat.clusterstatistic = 'maxsum';
cfg.stat.minnbchan = 2;
cfg.stat.tail = 0;
cfg.stat.clustertail = 0;
cfg.stat.alpha = 0.025;
cfg.stat.numrandomization = 500;
cfg.stat.computecritval = 'yes' ; % calculates the critical values of the test statistics
cfg.stat.computeprob = 'yes' ; %calculates the p-values
cfg.stat.frequency = 'all';
cfg.stat.ivar = 1; %conditions
cfg.stat.uvar = 2; %Participnts

I generated and saved the results as stats_results.mat.

The output yielded the following fields:
Field
stat
df
critval
prob
mask
dimord
dim
inside
pos
cfg

Tutorials I saw on the ft website included fields like posclusters angled negclusters when using Montecarlo method.

Thank you very much.

Best,
Monalisa




_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fmailman%2Flistinfo%2Ffieldtrip&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C9c4ab2b947b347f889a008dc546dd5da%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638478077945050349%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=wnLUzMVACQZArEJRKvryu3GFIeClxkHQ8H9adWyicSk%3D&reserved=0>
https://doi.org/10.1371/journal.pcbi.1002202

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240404/d631d64f/attachment.htm>