Dear FSL Experts,
I would like to explore correlation (positive and/or negative) between my FA images and a clinical measure. I have demeaned my measure (subtract the mean from measure of every subject) and copied and pasted using the fsl glm. I set the contrast to be 1 and -1, and have the following:
matrix (for simplicity, I made up the matrix):
/NumWaves 1
/NumPoints 30
/PPheights 6.1e+01
/Matrix
-1.4
-1.2
3.4
...
-6.5
contrast
/ContrastName1 poscorr
/ContrastName2 negcorr
/NumWaves 1
NumContrasts 2
/PPheights 6.1e+01 6.1e+01
/RequiredEffect 3.678 3.678
/Matrix
1
-1
I then run the following command:
randomise -i FA -o FA_stat -m FA_mask -d clinical.mat -t clinical.con -n 500 --T2 -V -D
I got the following warning: You have demeaned your data, but at least one design column has non-zero mean.
Since when contracting individual measures from the mean of the group, I need to round up the numbers when copying and pasting, is that why I got this warning?
Can I simply put the raw clinical data (no demean) into the matrix, and run the command without the -D option? Is that the same thing?
Thank you for your response in advance.
|