ft_freqstatistics for EEG data (cluster correction)

[DCP-INS]

Hi,
I would like to use the Fieldtrip function ft_freqstatistics to perform cluster statistics at the channel level on frequency-resolved (but not time-resolved) EEG data (63 channels), comparing 2 conditions. For each participant and condition the data have the following size: 63x2x20 (channels/time/frequency). When I run the process, even with only 2 participants (and 100 permutations), I obtain after 2 minutes a Java exception (OutOfMemoryError).
I run the process on a mac book pro 2016 (3,1 GHz Intel Core i7). I have a java heap size of 4,096MB on MATLAB.
This exception does not occur if I use another type of correction (bonferroni, fdr...).

Might you please help me to resolve this issue ?
Please find enclosed the input data (cfg = statcfg, and the 2*2 (2 participants * 2 conditions) data = ftAllFiles): thedata.mat.zip

thank you,
Benjamin Morillon

ps: the info appearing on the Matlab screen when I run the process:
the call to "ft_selectdata" took 0 seconds and required the additional allocation of an estimated 9 MB
using "ft_statistics_montecarlo" for the statistical testing
using "ft_statfun_depsamplesT" for the single-sample statistics
constructing randomized design
total number of measurements = 4
total number of variables = 2
number of independent variables = 1
number of unit variables = 1
number of within-cell variables = 0
number of control variables = 0
using a permutation resampling approach
repeated measurement in variable 2 over 2 levels
number of repeated measurements in each level is 2 2
computing a parametric threshold for clustering
computing statistic
estimated time per randomization is 0.07 seconds
computing statistic 100 from 100

[ftadel]

How are we supposed to reproduce your error?
If I load your file and run: ft_freqstatistics(statcfg, ftAllFiles{:}), it works with fieldtrip-20170809 on windows 10.
Which version of FieldTrip are you using?
Have you tried on a different computer?

[DCP-INS]

Thanks, François, for your reply. That's what I wanted to know (files are good, it should work as is). I'll thus download the last version (I had 20170401) and try on a different computer.
cheers

[DCP-INS]

Hi again,
After installing the last version, I obtain the following error, seemingly linked to the function spm_bwlabel (see below). Do you have by chance any idea of the underlying issue ?
thanks
bn

Error in findcluster (line 90)
[labelmat(spatdimlev, :, :), num] =
spm_bwlabel(double(reshape(onoff(spatdimlev, :, :), nfreq, ntime)), 6); % the
previous code contained a '4' for input

Error in clusterstat (line 197)
posclusobs = findcluster(reshape(postailobs,
[cfg.dim,1]),channeighbstructmat,cfg.minnbchan);

Error in ft_statistics_montecarlo (line 347)
[stat, cfg] = clusterstat(cfg, statrand, statobs);

Error in ft_freqstatistics (line 193)
[stat, cfg] = statmethod(cfg, dat, design);

[robertoostenveld]

which version of fieldtrip are you using (which date or which git version code), which version of MATLAB are you using and on which operating system are you running this?

[robertoostenveld]

and what is the exact (and full) error message that you see on screen?

[DCP-INS]

Hi,
Thanks for your reply.
I am using fieldtrip 20170914 on Matlab R2017a, on a macOS system (sierra 10.12.4).
The error message is so long that I can't grab it all. Basically it's associated with the compilation of mex files related to the function spm_bwlabel (called in findcluster.m, l.90). Attached is copy/pasted the portion of the error message: errormessage copy.docx

that I can retrieve.
Thank you,
Benjamin

[robertoostenveld]

I suppose you are running into something like this http://bugzilla.fieldtriptoolbox.org/show_bug.cgi?id=3330.

SPM mex files are in external and not maintained by the FieldTrip team. Please recompile the spm mex files for your platform or contact the SPM team.