Dear David,
On 25 Apr 2012, at 15:39, David Leitman wrote:
> Hi Vladimir,
> I have a a merged file for each subject that has multiple
> conditions, i created contrasts of these conditions using
> spm_eeg_weight_epochs , including a contrast reflecting the mean of
> all conditions.
>
> I then conduct a TF analysis using spm_eeg_avg_TF.
This sounds a little strange to me. There is no function with this
name in SPM. Furthermore, it only makes sense to compute contrasts on
evoked responses whereas it usually makes sense to do time-frequency
analysis of epoched data and then average. So I'm not sure your
pipeline results in something sensible.
>
> However I would then like to rescale all the created contrasts
> containing the different conditions to the contrast reflecting the
> mean of all conditions. Is there a way to do this young the SPM
> TFresacle function?
Do you mean that you want to divide one time-frequency matrix by
another? The only way to do it using GUI in SPM that I can think of is
to export your time-frequency data to images and then use ImCalc.
Otherwise, you'll have to write your own code (which is not that
complicated).
>
> The GUI only gives me the option to enter a second file.
>
> My second question is more basic ( or dumb) when conducting a
> wavelet analysis using spm_eeg_TF or spm_eeg_avg_TF, is it possible
> to select a range of wavelet cycles .
> so lets say I am interested in frequencies ranging from 1hz to 80 hz
> can I enter wavelet orders like 2:10 so that lower order( number of
> cycles) are used for lower frequencies and higher order wav lets are
> used for the higher frequencies?
>
The definition of wavelet decomposition is that you have fixed number
of cycles for all frequencies. You can in SPM fix the time window
instead but this is equivalent to doing fourier transform with the
hanning taper. If you want to have more flexible frequency resolution
which varies with frequency you can use the 'fieldtrip multitaper'
option.
Best,
Vladimir
> thanx David
|