Single-trial ITC

I’m doing some computations with tfr_multitaper where I need to access single-trial TFR data. This works fine with power by passing average=False and return_itc=False. This returns an EpochsTFR object, which contains the TFRs for each single trial.

However, setting return_itc=True (the default) in combination with average=False doesn’t work, because I get an error saying “Inter-trial coherence is not supported with average=False”. I think this is because ITC is only defined as an average over epochs, but I’m still looking to get the single-trial phase values. These must be available somewhere in the calculation process, but presumably they get averaged before they are returned as an AverageTFR object. Does anyone have an idea where these single-trials phase values are available?

Note that there seems to be an error in the docstring, because it says itc : AverageTFR | EpochsTFR, but apparently this cannot be an EpochsTFR object (this would instantly solve my problem).

we could add an output parameter to obtain the complex-valued Fourier coefs

Alex

I’ll see if I can find where these coefficients are available and will try to make a PR.

@cbrnr BTW, you can use output='complex', at least with tfr_morlet.

1 Like

Hello Clemens,

did this ever get implemented? I am also interested in single trial phase locking.

Best,

Carina

I think so: ENH: return complex multitaper output per taper by mmagnuski · Pull Request #10281 · mne-tools/mne-python · GitHub