Dear Stefan,
> I would like to compare Z-scores and extent of activation in fMRI data sets
> obtained with voxel sizes of 1.5x1.5x4 mm3, 3x3x4 mm3 and 6x6x4 mm3.
>
> Which smoothing parameters do I have to apply in order to obtain the SAME
> smootheness in ALL data sets (i.e. the high res. data sets must be smoothed
> more than the low res. data sets)?
An approximate solution would be to assume that the intrinsic FHWM of the
data was voxel size V. Therefore after convolution with a kernel with
FWHM F the effective FWHM E will be given by:
E = sqrt(V.^2 + F.^2)
if you require E to be the same E then
F = sqrt(E.^2 - V.^2)
> I noticed that the smoothness varies slightly between data sets that were
> obtained with the same spatial resolution and smoothing after motion
> correction was applied.
Removing movement-related components from the data will affect the
spatial correlation structure of the residuals because this component
may have a different correlation structure than the underlying error.
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|