Dear FSL experts,
I would like to smooth PET images for patients n=10 and controls n=10. Which command do you recommend:
fslmaths input.nii -kernel gauss 2.55 -fmean smoothed_output -odt float
fslmaths input.nii -s 2.55 -mas whole_brain_mask smoothed_output -odt float
The purpose of the question is :
I used randomise to run voxelwise analysis between the groups (10 patients vs 10 controls) using the smoothed PET images derived from the first command, then I repeated the same analysis (i.e. same randomise parameters) using the smoothed PET imaged derived from the second command. The power of the analysis (i.e. significant difference at lower p values) was better for the smoothed PET images derived from the second command line. Please what is the difference between the previous commands and why I see this difference in the results?