Hello Bernadette,
thank you for your answer.
Would the change in the size of the baseline have an effect on the
outcome of the DCM? If this is the case how could I define the
"right/best" interval for the baseline?
For example in our experiment we have a button press at 0ms pst and a
visual stimulus at approximately -400 to -300 ms pst. Would you
suggest to change the baseline interval to somewhere around <-500ms or
should we just use the predefined interval?
Best,
Nils
__________________
Dipl.-Math. Nils Rosjat
University of Cologne
Biocenter
Zuelpicher Str. 47b
50674 Cologne
[log in to unmask]
Zitat von Bernadette van Wijk <[log in to unmask]>:
> Hi Nils,
>
> The TF spectra express the relative log power for the peristimulus time
> window compared to baseline (which is the first 1/8 of samples):
>
> log(power) - log(average power baseline interval).
>
> Hence 0 represents no change in power compared to baseline.
>
> The baseline can be adjusted by manually changing the code at line 353 in
> spm_dcm_ind_data.m: n=fix(Nb/8).
>
> The log transform is taken in line 339: Y(f,k)=log(y(:));
> The baseline is subtracted in line 357: Yz{i,j} = Yk -Yb;
>
>
> Best,
> Bernadette
>
>
> 2014-10-10 23:37 GMT+01:00 Nils Rosjat <[log in to unmask]>:
>
>> Dear SPM users,
>>
>> in DCM for induced responses the first step is a wavelet transformation on
>> each trial of the data. The TF-trials are then averaged to get induced
>> time-frequency that is afterwards. I have two questions:
>>
>> 1.) What is the unit of the power shown in time-frequency result?
>> (dB,%,...)
>>
>> 2.) Is the time-frequency rescaled in some way, like relative or log? If
>> this is the case which time interval is used as baseline and is it possible
>> to choose this interval (if used at all) manually?
>>
>> Best regards,
>> Nils
>> __________________
>> Dipl.-Math. Nils Rosjat
>> University of Cologne
>> Biocenter
>> Zuelpicher Str. 47b
>> 50674 Cologne
>> [log in to unmask]
>>
>
|