[FieldTrip] 回复： On meaning of numbers in posclusterslabelmat / negclusterslablemat in results of ft_freqstatistics function

[郭策2021220122]

Thank you for your explanation. You mentioned "the stat.posclusterslabelmat will tell you the time and frequency extent of the cluster", so further I want to know are all integers in stat.posclusterslabelmat which are > 1 suggest in THIS channel at THIS frequency and at THIS timepoint, （based on my TF data calculated with slide length length, baseline and other parameters I set or factors in my data）, the p-value is lower than which I set in cfg.alpha when conducting permutation?



---- 回复的原邮件 ----
| 发件人 | Julian Keil via fieldtrip<fieldtrip at science.ru.nl> |
| 日期 | 2024年05月22日 13:20 |
| 收件人 | FieldTrip discussion list<fieldtrip at science.ru.nl> |
| 抄送至 | Julian Keil<julian.keil at gmail.com> |
| 主题 | Re: [FieldTrip] On meaning of numbers in posclusterslabelmat / negclusterslablemat in results of ft_freqstatistics function |
Hi,


these are the clusters identified in your data. The associated statistics can be found in stat.posclusters.
For example, stat.posclusters(1) will give you the information on the first cluster, and the stat.posclusterslabelmat will tell you the time and frequency extent of the cluster. Please see this post https://www.fieldtriptoolbox.org/faq/how_not_to_interpret_results_from_a_cluster-based_permutation_test/ for important information regarding the interpretation of the cluster stats results.


Best,


Julian



Am 22.05.2024 um 05:19 schrieb 郭策2021220122 via fieldtrip <fieldtrip at science.ru.nl>:


Hi, 


After conduction of cluster based permutation using ft_freqstatistics, a stat variable can be generated, containing two structs, namely poscterlabelmat and negclusterslabelmat. Both are 3-D data containing integers. I wonder the what is the meaning of the integers because they're not 0-or-1. Instead there're integers like 3, 5, 19, etc. How are these numbers determined? Do all non-zero positive integers indicate "this is a statistically significant point (p<=cfg.alpha)"? Are they determined by p-value?
The picture below can be generated with script
    contourf(0:4:999,frex,squeeze(stat.negclusterslabelmat(22,:,101:350)),50,'linecolor','none');
    set(gca,'clim',[-3 3],'xlim',[0 996]);
xlabel('Time (s)'), ylabel('Frequency (Hz)');
(Note that my TF cluster based permutation was conducted with TF data with a sampling rate of 250 and epoch ends [-0.4s 1.4s], so the script was written to draw a "map" of significant negative clusters within [0s 1s] in TF data comparison, of channel No.22 which was Cz. )
<Neg_2Pri_MalePrimed_&_FemalePrimed.jpg>
As you can see, there are different depths of color because of different values of non-zero positive integers. Should I say this map shows significant negative TF clusters?
 
Furthermore, what are differences between positve clusters and negative clusters? 
For example, if I have 2 variables to be tested, namely TFRwave1 and TFRwave2, does postive clusters generated with
     [stat] = ft_freqstatistics(cfg, TFRwave1, TFRwave2);
equal negative clusters generated with
    [stat] = ft_freqstatistics(cfg, TFRwave2, TFRwave1);
?


_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240522/445d2e98/attachment.htm>