Dear Amir,
You can localize low frequencies with short time windows. The only
problem is the frequency resolution. As we discussed some time ago,
the frequency resolution is 1/(time window). In particular, the first
frequency bin averages over all frequencies from DC to 1/(time
window). So with very short time window, let's say 100 ms you will be
able to look at 0 to 10 Hz including 4 Hz just won't be able to
separate 4 Hz from other frequencies below 10 Hz. With 400 ms window
you have 2.5 Hz frequency resolution. So you'll have bins for 0-2.5
and 2.5-5 Hz. You can also center your bin at 4Hz, but then it'll be
just a kind of interpolation of those optimal bins. In any case the
frequency resolution will not change.
For methods where there is no explicit frequency window (e.g. LCMV)
you can pre-filter your data around the frequency range of interest.
Best,
Vladimir
On Thu, Jan 28, 2010 at 3:15 AM, Amir H Javadi <[log in to unmask]> wrote:
> Hi :-)
> I have a question about the minimum time interval for different frequency
> bands for source localisation using different methods. I tried different
> types of source localisation and except "Fieldtrip beamformer source
> localisation", the 3rd option in "Beamforming..." method, all of them allow
> source localisation for short time intervals for low frequency bands. Well,
> in some of the methods, those in "3D source reconstruction" method, you are
> not asked for specifying frequency band. The other two source localisation
> methods, beamforming methods, "volumetric LCMV beamformer" and "Fieldtrip
> DICS beamformer" methods, you need to specify the frequency band explicitly.
> I wonder for example a 400ms time interval is long enough for source
> localisation of 4Hz brain activity. Very many thanks.
> Have a good time :-)
> Amir
>
>
>
|