[FieldTrip] Sharp 60 Hz peak (in Europe!) in wPLI and Granger(fieldtrip Digest, Vol 96, Issue 4)

[Antoine Ducorps]

Dear Eelke,

I agree that a sharp 60 Hz peak is certainly an artefact.
You mention that your projector is set to 120 Hz, but unless you use a very sophisticated one, the internal rate for most (if not all) commercial devices is still 60 Hz, and any input rate will be interpolated to that fundamental.
In your case, every second frame would be displayed, or an average light value of two frames, I am not sure.
You can check that on your technical manual, or alternating black/white frames and measuring the light output with a photodiode.
If it really happens that your projector runs internally at 60 Hz, then you should also set your computer output at the same value, otherwise you don’t know what the interpolation or down sampling will do, most probably not documented by the vendor.
Then, either you have a shielding problem with your projector electronics, or the 60 Hz rate is a real light effect in your stimulus, reflected in the visual cortex.

Best,

	Antoine.

> ------------------------------
> 
> Message: 3
> Date: Mon, 5 Nov 2018 17:21:05 +0100
> From: Eelke Spaak <e.spaak at donders.ru.nl>
> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
> Subject: [FieldTrip] Sharp 60 Hz peak (in Europe!) in wPLI and Granger
> 	spectra
> Message-ID:
> 	<CABPNLUryO7E-TnhKPwDWp3_i4GG6FJM2NukqVwDOLb+5hjCOgg at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
> 
> Fellow FieldTrippers,
> 
> I have identified two points of interest in the brain, between which I
> want to compute connectivity metrics. One is in early visual cortex,
> the other is in right temporal lobe, somewhat medial (compatible with
> hippocampus, but exact interpretation not relevant right now). I
> reconstructed activity for these grid points using LCMV beamformer (on
> MEG data) and computed source-level Fourier spectra (taper = 'dpss',
> tapsmofrq = 3) after applying a band-stop filter around 50 Hz and
> harmonics. Using ft_connectivityanalysis, I computed both debiased
> wPLI and Granger causality between the two points of interest.
> 
> In both spectra, I see a clear peak at 60 Hz in the grand average
> across 36 subjects, which is also there in the majority of individual
> subjects (though very strong only in 2/36). Plots are attached. Now
> this would be exciting news if indeed it turns out to be a true highly
> band-limited gamma effect!
> 
> But of course I suspect that this peak could very well be artifactual.
> I can't think of any artifact source I may have missed that would
> cause this, though. The projector refresh rate during the experiment
> was set to 120 Hz, and not 60. (Also note this is European data so AC
> frequency is 50 Hz and not 60, hence the band stop I mentioned
> earlier.)
> 
> Does anyone have any idea what might be causing this sharp peak? (A
> genuine effect after all?)
> 
> Thanks,
> Eelke
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: granger-spectrum.png
> Type: image/png
> Size: 14990 bytes
> Desc: not available
> URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20181105/4a5ee49b/attachment-0002.png>
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: wpli-spectrum.png
> Type: image/png
> Size: 11765 bytes
> Desc: not available
> URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20181105/4a5ee49b/attachment-0003.png>
> 
> ------------------------------
> 
> Message: 4
> Date: Mon, 5 Nov 2018 17:13:05 +0000
> From: "Schoffelen, J.M. (Jan Mathijs)" <jan.schoffelen at donders.ru.nl>
> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
> Subject: Re: [FieldTrip] Sharp 60 Hz peak (in Europe!) in wPLI and
> 	Granger spectra
> Message-ID: <ABDA30A7-342C-4CFE-9854-F2CEA9961B8B at donders.ru.nl>
> Content-Type: text/plain; charset="us-ascii"
> 
> This looks like an artifact to me.
> 
> In addition, the zigzag in the Granger spectra is suggestive of convergence issues in the non-parametric spectral matrix factorization in at least a few of your subjects. This is often caused by strong discontinuities in the spectra (spectral lines!) or poor behaviour of the time domain data at the edges of your epochs, causing spectral leakage.
> 
> JM
> 
>