Dear SPMers,
I have a question regarding con_*.nii and spmT_*.nii in fMRI processing.
I apologize if this is an old issue that has been already addressed.
I found that contrast estimate values in voxels in con_*.nii were not
necessarily consistent with t-values in the corresponding spmT_*.nii.
I had naturally thought that the higher the t-values, the higher the
contrast estimate values also. In reality, it was not the case.
I would like to know the reason for this discrepancy.
By this discrepancy, I found the following.
(1) The peak voxel in con_*.nii may not be the peak voxel in the
corresponding spmT_*.nii.
(2) Suppose in a con_*.nii file, the voxel value of the voxel A would
be greater than that of the voxel B. The opposite direction would be
possibly observed in the corresponding spmT_*.nii file; the voxel
value of the voxel B would be greater than that of the voxel A.
(3) Suppose we obtained voxels above p<0.001 as well as those above
p<0.01. I found that the top value of contrast estimate for p<0.001
could be smaller than that for p<0.01. I previously expected the same
value for these, but it was not the case. Instead I found that the
voxels above the threshold of p<0.001 (threshold by using spmT_*.nii)
did not include the top value of entire contrast estimates of the con
file.
I hope I could have sufficiently explained my question. Thank you very
much for your kind explanation.
Best regards,
Kayako
|