[Eeglablist] newtimef with large frequency windows

[Bradley Voytek]

Zach:

It's not so much that the parameters are forcing an 84 ms window, as
they're forcing 558 ms edges. You need to increase your epoch length
or--if I remember newtimef correctly--you might be able to increase
your padratio to change the size of these windows.

There are ways to get the information through bandpass filtering
though: you can get the analytic amplitude of the bandpassed signal.

::brad

On Mon, Aug 11, 2008 at 6:53 AM, Zachary Moran <zdmoran at gmail.com> wrote:
> Hi Brad,
>
> Thanks much for the response!  It will be helpful to know the exact window
> size required for a successful decomposition.
>
> As for why I'm using time-freq decompositions rather than a band pass
> filter, it's because I'm interested in exploring the degree to which many
> different frequency bins contribute to event-related waveforms rather than
> just paring down the broadband signal.  As far as I'm aware, I don't think I
> could do this through filtering, though if you know of a method I'd greatly
> appreciate any help or references.  Also, I think wavelet transforms differ
> from filters in terms of frequency resolution, roll-off/leakage especially
> in the lower frequencies.
>
> My big problem has been that the ERSP array that I get seems to be stuck
> from 358.0 to 442.0 ms whereas I want to select elements of the array
> corresponding to early, evoked potentials.  I think the decomposition will
> enforce that this window be 84 ms long given the other dataset parameters,
> but does anyone know if there is some way to recenter the window for a
> different 84ms (e.g., 66 to 150)?
>
> Thank you!!
>
> Zach
>
> On Sun, Aug 10, 2008 at 3:31 PM, Bradley Voytek <bradley.voytek at gmail.com>
> wrote:
>>
>> Zach:
>>
>> Your decomposition parameters require a 558 ms windows on either side
>> of your epoch, so if you want a 200 ms baseline you'll have to have a
>> minimun time of your epoch be (200 + 558) = 758 ms. Your ending time
>> of 1000ms should be sufficient for the timing of the ERP components of
>> interest.
>>
>> That said, with such a large pass band, why not just band pass your
>> raw signal between 3 and 60 Hz? Why bother with a time-frequency
>> decomposition?
>>
>> ::brad
>>
>>
>> On Sat, Aug 9, 2008 at 2:57 PM, Zachary Moran <zdmoran at gmail.com> wrote:
>> > Hi all,
>> >
>> >
>> > At the moment I'm working on a project in which I'm trying to get ERSP
>> > estimates around some early evoked potentials (e.g., between 20 and 100
>> > ms).  Though I'd like to implement a frequency window of 3 to 60 Hz,
>> > when I
>> > attempt to do so, EEGLAB forces the output to fall within a window of
>> > 358.0
>> > to 442.0 ms.  Unfortunately, the only way I'm able to get to earlier
>> > time-points is to increase my minimum frequency to around 12 Hz thus
>> > allowing for an overall larger time window.  However, I have seen the
>> > sort
>> > of analysis that I'm looking for done (I think in EEGLAB, since it used
>> > ERSP
>> > and ITC specifically) by Brockhaus-Dumke et al. (2008)  "Sensory gating
>> > revisited: Relation between brain oscillations and auditory evoked
>> > potentials in schizophrenia."  I've included an example of my script
>> > below:
>> >
>> > <<figure; pop_newtimef( EEG, 1, 1, [-200  1000], [3         0.5]
>> > ,'type',
>> > 'phasecoher', 'topovec', 1, 'elocs', EEG.chanlocs, 'chaninfo',
>> > EEG.chaninfo,
>> > 'title','ERSP and ITC','padratio', 1, 'plotphase','off','freqs',[3
>> > 60]);>>
>> >
>> > My dataset is sampled at 1000 Hz and has 1201 frames per epoch.  Does
>> > anyone
>> > have any experience with this who knows how I might get around this
>> > problem
>> > or any way of recentering the output window around desired ERP
>> > components
>> > (e.g., from 75 to 125)?  Any help or information would be very greatly
>> > appreciated.
>> >
>> >
>> > Thank you very much,
>> >
>> >
>> > Zach
>> >
>> >
>> > Zachary Moran
>> > Graduate Student
>> > University of California-Los Angeles
>> >
>> > _______________________________________________
>> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> > To unsubscribe, send an empty email to
>> > eeglablist-unsubscribe at sccn.ucsd.edu
>> > For digest mode, send an email with the subject "set digest mime" to
>> > eeglablist-request at sccn.ucsd.edu
>> >
>
>