[FieldTrip] NaNs in power spectrum output from ft_freqanalysis

Sara Pizzamiglio S.Pizzamiglio at uel.ac.uk
Tue Jul 7 21:00:05 CEST 2015

Dear FieldTrippers,

I am facing a problem with ft_freqanalysis which has already been reported here in the mailing list but not clear and definite answer/solution has been given so far.

I am working with resting-state eeg, 3 minutes eyes closed which I divide in epochs of 1 sec each, I preprocess and carefully visually inspect to discard any artefact. When channels are noisy I use ft_repairchannel with the method triangulation to repair them. I am focusing only on the right and left motor areas, plus the middle line electrodes.

I would like to analyse the signals in the time-frequency domain, thus once the data are all preprocessed and cleaned (contained in the structure DataOut_Rep) I perform an analysis with Morlet wavelet. Below there is my code:

cfg = [];
cfg.channel = {'all' '-HEOG' 'VEOG'};
cfg.foi = 3:1:45;
cfg.toi = 0:0.001:1;
cfg.output = 'pow';
cfg.method = 'wavelet';
cfg.width = 7;
cfg.t_ftimwin = ones(length(cfg.foi),1).*0.5;
TFR = ft_freqanalysis(cfg, DataOut_Rep);

Here is an example of dataset (this in particular is baseline, resting state before any task is performed):


I am facing 2 problems with the output of ft_freqanalysis:

1)      I have a lot of NaNs: TFR.powspctrm is made of 29 (channels) x 43 (frequencies) x 1001 (time points), and in several position I have NaN values.

2)      When I use ft_topoplotTFR, the plots of some frequency bands (e.g. Beta, 13-30 Hz) have super high values, up to hundreds or thousands.

Has anybody any idea of why this happens?! In case, does also anyone have a potential solution to overcome this issue?!

Thank you!
 This email has been scanned for email related threats and delivered safely by Mimecast.
 For more information please visit http://www.mimecast.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150707/e00578a0/attachment-0001.html>

More information about the fieldtrip mailing list