For removing the problem that different slices were acquired at different time, I tried three methods seperately:
1) Slicetiming.
2) Time and dispersion derivatives (without slicetiming).
3) Both slicetiming and time and dispersion derivatives.
In significance, generally, p2 < p3 < p1, showing that approach 2 is preferable in this aspect.
I'm now wondering: if i adopt approach 2, can we still compare PSTH (peri-stimulus histograms) of two regressors / two voxels (like the bottom part of figure 29.14 in p.240 in SPM8 manual)? (If no, what is the appropriate alternative way?) As we know, the time derivatives are different for different voxels and regressors. So PSTH for bf1 of each regressor or voxel is in fact shifted forward or backward in different time length. So when we compare PSTH for bf1 of two regressors or two voxels, even if the peak of regressor / voxel A appears earlier than that of regressor / voxel B, we don't know whether it is really so or it is just caused by the different time derivatives.
Another related question is: can we integrate dispersion derivative into PSTH? If yes, how? If no, what is the appropriate alternative way?
Looking forward to your help.
Thank you in advance,
James
|