[Eeglablist] wavelet properties
Makoto Miyakoshi
mmiyakoshi at ucsd.edu
Thu Sep 17 18:57:10 PDT 2015
Dear Hendrik,
> Using 3 cycles at lowest frequency to 12.8 at highest.
> Okay, so this means that the wavelet has 3 cycles at my lowest estimated
frequency and 12.8 cycles at the highest. If my frequency resolution based
on the 5th output argument from newtimef() shows me 46 frequencies with the
highest one being 50, then the 12.8 cycles correspond to the 50 Hz correct?
Yes, correct.
>Generating 200 time points (-715.0 to 1713.0 ms)
Finding closest points for time variable
Time values for time/freq decomposition is not perfectly uniformly
distributed
The window size used is 285 samples (570 ms) wide.
> I assume this is what you mean with the sliding window? Does this mean
the length of each wavelet in time is 570 ms?
Yes. And it slides 200 times to generates 200 time points in this case. The
sliding width is 1713-(-715) divided by 200. It makes sense, doesn't it?
> Estimating 46 linear-spaced frequencies from 5.9 Hz to 50.0 Hz.
> And I was wondering if this means that my time-freq decomposition is
based on 46 wavelets? I am just curious since we use wavelets for
electromyography processing in our lab and they are typically a lot wider
space so I just want to make sure that I get this right. I have 46 wavelets
to cover up to a frequency range of 50 Hz?
Yes, you generated 46 wavelet daughters covering 5.9 to 50.0 Hz. These
wavelets have both temporal and frequency distribution, so they will nicely
overlap in both domains.
Makoto
On Sun, Sep 6, 2015 at 12:34 PM, Hendrik Enders <endersh at ucalgary.ca> wrote:
> Hi Makoto,
>
>
>
> As always, thanks for your help and your quick reply. I ran the code on
> some data of my subjects to obtain the information. I copy the command line
> output just to check if I draw the right conclusions.
>
>
>
> Computing Event-Related Spectral Perturbation (ERSP) and Inter-Trial Phase
> Coherence (ITC) images based on 222 trials of 1500 frames sampled at 500 Hz.
>
> Each trial contains samples from -1000 ms before to 1998 ms after the
> timelocking event.
>
> Image frequency direction: normal
>
> *Makes all sense.*
>
>
>
> Using 3 cycles at lowest frequency to 12.8 at highest.
>
> *Okay, so this means that the wavelet has 3 cycles at my lowest estimated
> frequency and 12.8 cycles at the highest. If my frequency resolution based
> on the 5th output argument from newtimef() shows me 46 frequencies with the
> highest one being 50, then the 12.8 cycles correspond to the 50 Hz correct?*
>
>
>
> Generating 200 time points (-715.0 to 1713.0 ms)
>
> Finding closest points for time variable
>
> Time values for time/freq decomposition is not perfectly uniformly
> distributed
>
> The window size used is 285 samples (570 ms) wide.
>
> *I assume this is what you mean with the sliding window?
> Does this mean the length of each wavelet in time is 570 ms?*
>
>
>
> Estimating 46 linear-spaced frequencies from 5.9 Hz to 50.0 Hz.
>
> *And I was wondering if this means that my time-freq decomposition is
> based on 46 wavelets? I am just curious since we use wavelets for
> electromyography processing in our lab and they are typically a lot wider
> space so I just want to make sure that I get this right. I have 46 wavelets
> to cover up to a frequency range of 50 Hz?*
>
>
>
> Thanks for your help. As always, much appreciated.
>
> Hendrik
>
>
>
> *Hendrik Enders*
>
> *PhD Candidate*
>
> *NSERC Vanier scholar*
>
> Faculty of Kinesiology
>
> University of Calgary
>
> 2500 University Dr NW
>
> Calgary, AB, Canada, T2N 1N4
>
> T: (403) 220-2413
>
> E: endersh at ucalgary.ca <henders at kin.ucalgary.ca>
>
>
>
> *From:* Makoto Miyakoshi [mailto:mmiyakoshi at ucsd.edu]
> *Sent:* September-02-15 12:40 PM
> *To:* Hendrik Enders <endersh at ucalgary.ca>
> *Cc:* eeglablist at sccn.ucsd.edu
> *Subject:* Re: [Eeglablist] wavelet properties
>
>
>
> Dear Hendrik,
>
>
>
> When you run newtimef(),
>
>
>
> >> [ersp,itc,powbase,times,freqs,erspboot,itcboot] = ...
>
> newtimef(data, frames, epochlim, srate, cycles,...
>
> 'key1',value1, 'key2',value2, ... );
>
>
>
> The 5th output 'freqs' is the list of center frequencies. I think it uses
> 'freqresol' output from dftfilt3.
>
>
>
> > The second point that I need to find out is the duration in the time
> domain for each wavelet. Can anybody help how I can generate that
> information to provide a table with the wavelet information?
>
>
>
> When you run newtimef() it shows the length of the sliding window. You can
> report the length there.
>
>
>
> In EEGLAB's newtimef, by default the number of cycles increases linearly
> as the central frequency increases, which may be tricky to explain. The
> purpose of doing this is to adjust the ratio of time/frequency resolution
> along with the frequency axis (if you use the fixed cycle numbers the
> output image looks vertically stretched i.e. too much temporal resolution
> with too less freq resolution at high freqs)
>
>
>
> Makoto
>
>
>
> On Mon, Aug 31, 2015 at 5:05 PM, Hendrik Enders <endersh at ucalgary.ca>
> wrote:
>
> Hello,
>
>
>
> I have finished my time-frequency analysis of my EEG data. Some people
> have asked me for some very specific information regarding the wavelet
> properties and I am unsure how to retrieve this information. I have a
> sampling frequency of 500 Hz and I used the input [3 0.5] for cycles and a
> pad ratio of 2.
>
>
>
> How can I calculate the center frequency of each individual wavelet that
> was generated between 3 – 50 Hz?
>
> I have tried the following:
>
>
>
> freqs = [3 50];
>
> cycles = [3 0.5];
>
> srate = 500;
>
> [wavelet,cycles,freqresol,timeresol] = dftfilt3( freqs, cycles, srate);
>
>
>
> This only returns two cells with wavelet information, time resolution and
> frequency resolution. How can I use the function to obtain center
> frequencies?
>
>
>
> The second point that I need to find out is the duration in the time
> domain for each wavelet. Can anybody help how I can generate that
> information to provide a table with the wavelet information?
>
>
>
> Thank you very much.
>
> Hendrik
>
>
>
>
> _______________________________________________
> 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
>
>
>
>
>
> --
>
> Makoto Miyakoshi
> Swartz Center for Computational Neuroscience
> Institute for Neural Computation, University of California San Diego
>
--
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20150917/3f3a5275/attachment-0001.html>
More information about the eeglablist
mailing list