[Eeglablist] Single Trial phase locking measure

Mon May 8 14:07:59 PDT 2006

Dear Dr Ming Qian
My advice is to try the following method which I had already
designed and verified in my Ph.D.
see the paper at:
----------------------------------------------------------------------------
-----
      Acta Neurobiol Exp (Wars). 1995;55(3):177-91.

Variability analysis of visual evoked potentials in humans by pattern
recognition in phase domain.
Achimowicz JZ.
Department of CNS Diagnostic, Polish Air Force Institute of Aviation
Medicine, Warsaw, Poland.

A novel approach to single trial visually evoked potentials (VEP)
variability analysis based on a new model of post-stimulus brain electrical
activity is presented. The convolution model introduced by the author is
experimentally verified by the analysis of flash stimulus effects on EEG
amplitude and phase spectra. Pattern recognition in the signal phase domain
is proposed for detection of any time locked transient signals. This is
illustrated by an application of a clustering algorithm in two-dimensional
unwrapped phase of EEG Fourier transform space for occipitally recorded VEPs
in human subjects.

Publication Types: Clinical Trial PMID: 8553911 [PubMed - indexed for
MEDLINE]
-----------------------------------------------------------------
the method measures directly phase locking of ERP in case when there are
distinct peaks in power spectra of evoked responses at several (stable)
frequencies
it works like that:
1. Take consecutive single responses, remove DC, apply window with taper
starting before stimulus onset and ending
when there are no peaks in time domain which are of interest for you.
2. Do FFT on each of them.
3. Unwrap the phase of single ERPS FFT (procedure available in Matlab)
4. choose the frequencies of interest basin on average power spectrum of all
responses.
5. Evaluate statistically the randomness of phase at those frequencies using
e.g. Raylegh test (random phase has uniform distribution)

than you can state that phase of evoked response
at for example gamma frequency is
nonrandomly distributed = there is phase locking at this frequency with
given level of confidence
In my Ph.D. I had used multimentional (selected frequencies) phase feature
space to describe given set of single responses. Then I had clustered them
in that space and used cluster means as the measure of given type of
response like alfa triggered responce or stimulus specific response.
I will try to find electronic version of this paper and send it to you ASAP.

Jerzy  Achimowicz Ph.D. Dr.Sc.
IMGW-PO
Center for Development of Numerical Forecasts
Podlesna  61 str.
01-673 Warsaw, POLAND.
phone:  office:   +48(22) 569-4130
             mobile:  +48  514-834--544

----- Original Message -----
From: "Ming Qian" <qianmi at ee.duke.edu>
To: "Jerzy Achimowicz" <jerzy_achimowicz at imgw.pl>
Sent: Monday, May 08, 2006 8:39 PM
Subject: Re: [Eeglablist] Single Trial phase locking measure

> Dr Achimowicz:
>
> So your recommendation is that I should try to fit sines to a given
> single trial response.  I can assign a different frequency to each
> individual sine.  I can try 70 different sines using frequency from 1Hz to
> 70Hz.  And I can find out how fit of each sine (frequency) fit with my
> trial response.  Then what?  How am I going to measure the
> synchronization?
>
> Can you give me some more details?  Like what will you do?
>
>
> Thanks for the hint.
>
> Ming Qian
>
>
>
>
> On Sat, 6 May 2006, Jerzy Achimowicz wrote:
>
> > Dear Ming,
> > If I understsand properly ITC measures stability of delay=phase of
transient
> > signal in a defined time window.
> > If you have a single response the only thing you can calculate is the
dalay
> > or ERP versus time of window start
> > or stimulus onset time (time delay=phase*inverse of frequency of
component
> > at which you had calculated phase).
> > Of course the deley is not equal to any latency of the waveform. Its a
delay
> > of a given sinewave(frequency) fitted to single
> > response. To have the best fit you can insted of FFT use other
(statistical)
> > methods to best fit a couple of sines
> > to a given single trial response. This is the analog to SVD.
> > Hope I had made it clear to you.
> > Jerzy
> > PS. Ask more if yoy have more questions on the subject of phase
> > synchronization !
> >
> >
> > Jerzy  Achimowicz Ph.D. Dr.Sc.
> > IMGW-PO
> > Center for Development of Numerical Forecasts
> > Podlesna  61 str.
> > 01-673 Warsaw, POLAND.
> > phone:  office:   +48(22) 569-4130
> >              mobile:  +48  514-834--544
> >
> > ----- Original Message -----
> > From: "Ming Qian" <qianmi at ee.duke.edu>
> > To: "Scott Makeig" <smakeig at ucsd.edu>
> > Cc: <eeglablist at sccn.ucsd.edu>
> > Sent: Monday, May 01, 2006 11:02 PM
> > Subject: [Eeglablist] Single Trial phase locking measure
> >
> >
> > > Dear all:
> > >
> > > Inter-trial Coherence(ITC) is a statistical measure of the randomness
of
> > > phase at each latency, across trials.  I wonder what I should use if I
> > > want to measure the phase locking in a single trial?
> > >
> > > Thanks for any input you might have.
> > >
> > > Ming Qian
> > >
> > > _______________________________________________
> > > eeglablist mailing list eeglablist at sccn.ucsd.edu
> > > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > > To unsubscribe, send an empty email to
> > eeglablist-unsubscribe at sccn.ucsd.edu
> > >
> >
> >
>
>