<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Hi Nathan, <br>
    </p>
    <p>I completely agree with what Jan-Mathijs wrote. Just a simple
      addition to improve performance: You can build your decoder or
      Fourier amplitudes with harmonics but you can alternatively use
      CCA with a sines and cosines at the appropriate frequencies  as
      reference signals to transform your time series EEG signal. With
      MEG and EEG we and others found clear performance increases for
      CCA over FT (see reference below). CCA is to my knowledge
      available in fieldtrip.</p>
    <p>Best, <br>
    </p>
    <p>    Jochem <br>
    </p>
    <p>Hakvoort, G., Reuderink, B., and Obbink, M. (2011). Comparison of
      PSDA
<br>
      and CCA Detection Methods in a SSVEP-Based BCI-System. (CTIT
      Technical
<br>
      Report Series; No. TR-CTIT-11-03) Enschede: Centre for Telematics
      and
<br>
      Information Technology University of Twente.</p>
    <p>Reichert, C., Kennel, M., Kruse, R., Hinrichs, H., & Rieger,
      J. W. (2013). Efficiency of SSVEF Recognition from the
      Magnetoencephalogram. In Proceedings of the International Congress
      on Neurotechnology, Electronics and Informatics (BrainRehab-2013),
      pages 233-237, DOI: 10.5220/0004645602330237 <br>
    </p>
    <p><br>
    </p>
    <blockquote type="cite"
cite="mid:CAAKLYiiOW237sUp5_uvV-yWZUO9VOfgDbh+QcvxGpEVFMLk2uA@mail.gmail.com">
      <div>
        <div dir="ltr">
          <div class="gmail_quote">
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
              0.8ex;border-left:1px solid
              rgb(204,204,204);padding-left:1ex">
              <br>
----------------------------------------------------------------------<br>
              <br>
              Message: 1<br>
              Date: Thu, 23 Feb 2023 11:12:43 +0000<br>
              From: "Schoffelen, J.M. (Jan Mathijs)"<br>
                      <<a
                href="mailto:janmathijs.schoffelen@donders.ru.nl"
                target="_blank" moz-do-not-send="true">janmathijs.schoffelen@donders.ru.nl</a>><br>
              To: FieldTrip discussion list <<a
                href="mailto:fieldtrip@science.ru.nl" target="_blank"
                moz-do-not-send="true">fieldtrip@science.ru.nl</a>><br>
              Subject: Re: [FieldTrip] [Fieldtrip] Analysis Options for
              SSVEPs<br>
              Message-ID: <<a
                href="mailto:6D6ED6A7-9FC7-433E-AC7D-BD34E613C54F@donders.ru.nl"
                target="_blank" moz-do-not-send="true">6D6ED6A7-9FC7-433E-AC7D-BD34E613C54F@donders.ru.nl</a>><br>
              Content-Type: text/plain; charset="utf-8"<br>
              <br>
              Hi Nathan,<br>
              <br>
              This question is not at all dumb, it’s good one!<br>
              <br>
              A classical SSVEP analysis indeed estimates the power of
              the evoked response (i.e. first average the trials, then
              compute the power spectrum). This works well under the
              valid assumption that the timing of the individual stimuli
              is 1) very precise, and 2) not jittered across trials.<br>
              <br>
              Now, if you want to estimate time-varying changes in the
              evoked power - e.g. to use it as a ‘marker’ for the
              attentional focus - you need to keep in mind that the
              specificity of your spectral estimate (+your sensitivity
              to detect changes over time)  (i.e. the extent to which
              the estimate at - say - 15 Hz reflects only the energy in
              the signal at 15 Hz or that it also contains  energy of
              close by frequencies -> a phenomenon that relates to
              the notion of spectral leakage, and the time-frequency
              trade off) depends on your experimental stimuli and your
              analysis parameters. In your example, using frequencies of
              13 and 15 Hz, they may not be sufficiently far apart in
              the spectral domain in order to be sufficiently reliably
              resolved given your most likely choice sliding time window
              that you are going to use for your analysis.<br>
              <br>
              Concretely, the kernel that you will use for your
              time-frequency analysis should have a bandwidth that is
              small enough to be able to separate the stimulus
              frequencies. This requires a sliding time window width of
              at least one second (because a time window of 1 second
              yields a spectral resolution of 1 Hz), which is probably
              too long for your experimental design (because it would
              require trials that are correspondingly much longer than
              that time window obviously). Therefore, I suspect that you
              may need to think a bit (and discuss with your co-workers)
              the optimal stimulation frequencies for both hemispheres.<br>
              <br>
              Another thing to keep in mind w.r.t. the optimal
              frequencies, is that ideally the lower order harmonic
              frequencies of the stimuli don’t overlap, because it may
              be interesting later on to look at the power of the
              harmonics as well, and in this case those harmonics will
              still distinguish between stimuli.<br>
              <br>
              Good luck with your experiment,<br>
              <br>
              Jan-Mathijs<br>
              <br>
              <br>
              > On 23 Feb 2023, at 10:35, Nathan Han via fieldtrip
              <<a href="mailto:fieldtrip@science.ru.nl"
                target="_blank" moz-do-not-send="true">fieldtrip@science.ru.nl</a>>
              wrote:<br>
              > <br>
              > Hi all,<br>
              > <br>
              > I'm still quite a novice with Fieldtrip and signal
              processing so please forgive me if this question is dumb
              :)
              <br>
              > <br>
              > I'm running an experiment where there is a visual
              stimulus on the left and right side of the screens, the
              purpose of which is to elicit SSVEPs. One circle flickers
              at 13Hz (e.g., the left side) and the other flickers at
              15Hz (e.g., the right side). I would like to analyse the
              change in visual spatial attention over the course of the
              trial and one way I was thinking was that if attention
              switches from the left to the right side of the screen,
              that 13Hz power would reduce while 15Hz power would
              increase over the course of the trial.<br>
              > <br>
              > I'm not sure if that even makes sense or if it is
              possible. If it does, I would like to ask how should I
              approach this analysis?<br>
              > <br>
              > Kind regards,<br>
              > Nathan<br>
              > _______________________________________________<br>
              > fieldtrip mailing list<br>
              > <a
                href="https://mailman.science.ru.nl/mailman/listinfo/fieldtrip"
                rel="noreferrer" target="_blank" moz-do-not-send="true">
                https://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
              > <a
href="https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!4kaPlXy0aQBSGhIki7DsF5G65GY1RIR1O6zJS_9bZiMo7YBKwWNPv3iv94qjE9mOPGJoznVOqvchq0Flq9w_rEtP64v_t5LPvqi0Aw$"
                rel="noreferrer" target="_blank" moz-do-not-send="true">
https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!4kaPlXy0aQBSGhIki7DsF5G65GY1RIR1O6zJS_9bZiMo7YBKwWNPv3iv94qjE9mOPGJoznVOqvchq0Flq9w_rEtP64v_t5LPvqi0Aw$</a>
              <br>
              <br>
              <br>
              <br>
              <br>
              ------------------------------<br>
              <br>
              Subject: Digest Footer<br>
              <br>
              _______________________________________________<br>
              fieldtrip mailing list<br>
              <a
                href="https://mailman.science.ru.nl/mailman/listinfo/fieldtrip"
                rel="noreferrer" target="_blank" moz-do-not-send="true">https://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
              _______________________________________________<br>
              fieldtrip mailing list<br>
              <a
                href="https://mailman.science.ru.nl/mailman/listinfo/fieldtrip"
                rel="noreferrer" target="_blank" moz-do-not-send="true">https://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
              <a
href="https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!_DurIXGhhon-3u1rByXBHmnI4SC9tIwgPxxfFWaECrVU6QQwpNQRx3NqZueShw4UvNeDSYoTB51zcId5hMz9dq6RAzO4M_0$"
                rel="noreferrer" target="_blank" moz-do-not-send="true">https://doi.org/10.1371/journal.pcbi.1002202</a><br>
              <br>
              <br>
              ------------------------------<br>
              <br>
              End of fieldtrip Digest, Vol 147, Issue 19<br>
              ******************************************<br>
            </blockquote>
          </div>
        </div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
fieldtrip mailing list
<a class="moz-txt-link-freetext" href="https://mailman.science.ru.nl/mailman/listinfo/fieldtrip">https://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a>
<a class="moz-txt-link-freetext" href="https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!9XsKRtYOa3VHm9M-Ngz2Af8OeDKeP0eIhnDpQM2aYxcUh4g_FuRXGWwbkRawJDW2z2kaNsbGa3GUFnn3p7PeuNWJKWwGxdpo5IJO$">https://doi.org/10.1371/journal.pcbi.1002202</a>
</pre>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 


</pre>
  </body>
</html>