[FieldTrip] TFR plotting following ft_freqbaseline

[Herring, J.D. (Jim)]

Dear Paul,

You can check whether the baseline correction worked by seeing whether the mean in the baseline per frequency is zero. If your frequency data is contained in the structure ‘freq’ you could do something like the following:

baseline = [-.5 0];
channel_idx = find(strcmp(freq.label,’MLO11’));
time_idx = find(freq.time>=baseline(1)&freq.time<=baseline(2));

mean_baseline = nanmean(freq.powspctrm(channel_idx,:,time_idx),3);

‘mean_baseline’ should be roughly around zero for all frequencies. I suspect that the positive activity at/slightly before zero cancels the negative activity around -500 to -400. Here it is important to realize that whenever you are calculating a time-frequency analysis you are moving a sliding time-window over your data. The value at time-point zero therefore represent the power at that particular frequency from minus to plus half of the chosen window length. Given that your time-window is 500ms, the data at time-point zero therefore reflects the power from [-250 – 250ms]. It is therefore very likely that it contains stimulus-related activity. If you want to be sure that your baseline data does not represent any stimulus-related activity you should choose a baseline period earlier than, or at -250 ms (ie. Half your sliding time window size).

Best,

Jim


From: fieldtrip-bounces at science.ru.nl [mailto:fieldtrip-bounces at science.ru.nl] On Behalf Of Paul Metzak
Sent: Thursday, April 07, 2016 1:48 AM
To: fieldtrip at science.ru.nl
Subject: [FieldTrip] TFR plotting following ft_freqbaseline

Hello Fieldtrippers,

I have a quick question about the output from my ft_freqanalysis. I ran ft_freqbaseline but when I plot the TFR using TFRmultiplot, I still have activity in the baseline period. I have included the cfg for the ft_freqanalysis call as well as the cfg for the ft_freqbaseline call and a .png from one of the sensors with this message. I took a look at the TFR plots for the pre- and post- ft_freqbaseline call and they look different so it is doing it's job but I am just wondering if activity in the baseline period represents a problem with the way I have analyzed my data thus far, or whether it is a normal result in fairly rapid MEG designs.

Thanks for any information you can offer!

Paul


%%% ft_freqanalysis cfg %%%%


         tstep: 0.0500
       twindow: 0.5000
          fLow: 2
         fHigh: 160
         fstep: 2
        method: 'mtmconvol'
         taper: 'hanning'
         rmvbl: 'yes'
        bltype: 'absolute'
      baseline: [-0.5000 0]
      trlbegin: -1
        trlend: 2
    conditions: [1 2 3]
        output: 'powandcsd'
       runname: 'run1'
           foi: [1x80 double]
     t_ftimwin: [80x1 double]
           toi: [1x51 double]

%%% ft_freqbaseline cfg %%%%

        baseline: [-0.5000 0]
    baselinetype: 'absolute'


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160407/ae0828af/attachment-0002.html>