[FieldTrip] input data structure ft_freqstatistics

Schoffelen, J.M. (Jan Mathijs) jan.schoffelen at donders.ru.nl
Wed May 31 08:58:12 CEST 2017

Hi Luisa,

The minimum requirement for a data structure that fieldtrip can work with should contain at least the following fields:

- a field with the data matrix, in your case it can be cohspctrm
- a bunch of fields that indicate how to interpret the dimensions of the input data matrix:
  *  dimord, a descriptor in the style of ‘chancmb_freq_time’
  *  freq
  *  time
  *  labelcmb (for univariate data, it should be label)

The cfg should be optional.

Anyway, this being said, probably you may want to try and fool fieldtrip into getting a univariate data-field, i.e. rename cohspctrm into powspctrm, and labelcmb into label. This should work under the assumption that you are cohering 8 different EEG electrodes with a single EMG (for the label you could just take the column of labelcmb that represents the EEG).

Best wishes,


J.M.Schoffelen, MD PhD
Senior Researcher, VIDI-fellow - PI, language in interaction
Telephone: +31-24-3614793
Physical location: room 00.028
Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands

On 31 May 2017, at 04:28, Luisa Roeder <l.roeder at qut.edu.au<mailto:l.roeder at qut.edu.au>> wrote:

Dear Fieldtrip users,

I am part of a research team at the Institute of Health and Biomedical Innovation at Queensland University of Technology in Brisbane, Australia. I am currently working on time-frequency analyses of EEG and EMG signals.

I would like to use ft_freqstatistics for a cluster-based permutation test on my time-frequency data.
In the function itself and in the online documentation I read that the input data into ft_freqstatistics should be the result/output from either FT_FREQANALYSIS, FT_FREQDESCRIPTIVES or from FT_FREQGRANDAVERAGE.
I have not completed my time-frequency analysis via fieldtrip; hence, I am wondering if it is also possible to create my own input data structure to make it fit for ft_freqstatistics?
(Basically, I would like to skip ft_freqanalysis and directly start with cluster-based permutation testing via ft_freqstatistics, and for that I need to adapt my data structure.)  If it is possible, which fields have to be created in my variable to be fed into ft_freqstatistics, so it fulfils all requirements to make the function run?
Are there any fields from the output of, for example, FT_FREQANALYSIS that are not needed for ft_freqstatistics?

For example,
[stat] = ft_freqstatistics(cfg, coh1, coh2, ...)

What fields do coh1, coh2, … have to have? E.g.

>> display(coh1)
     labelcmb: {8×2 cell}
       dimord: 'chancmb_freq_time'
    cohspctrm: [8×80×2049 double]
         time: [1×2049 double]
         freq: [1×80 double]
         elec: [1×1 struct]
          cfg: [1×1 struct]

Can I leave out some of these fields (particularly coh1.cfg), and it will still work for ft_freqstatistics?
Would it be enough to only have coh1.cohspctrm, coh1.time and coh1.freq?

I am still relatively new to Fieldtrip, and working my way through the tutorials etc., but I definitely having trouble to create an appropriate data structure as input into ft_freqstatistics (without using ft_freqanalysis or other functions alike beforehand). I would very much appreciate your help with this.

Thank you,
Kind regards

Luisa Roeder, PhD | Research Associate | Movement Neuroscience
Institute of Health and Biomedical Innovation | School of Exercise and Nutrition Science
Queensland University of Technology | 60 Musk Ave, Kelvin Grove, Brisbane, QLD 4059, Australia.
t: +61 7 3138 6428 | e: l.roeder at qut.edu.au<mailto:l.roeder at qut.edu.au> | w: www.ihbi.qut.edu.au<http://www.ihbi.qut.edu.au/>
CRICOS No. 00213J

fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20170531/2d1628b0/attachment-0002.html>

More information about the fieldtrip mailing list