Dear friends,
I have a BOLD timecourse after preprocessing stages. Now I would like
to regress the motion confounds using the text file output from motion
correction. At the end, I need to get my BOLD timecourse without the
motion effects. So, I ran the 1-st level analysis with only motion
parameters, which resulted in 7 beta values per voxel. Now I assume,
that my motion estimated timecourse should be Y_motion =
X_motion_design_martix * Beta_matrix. Finally, to get the cleaned
timecourse I subtract Y_cleaned = Y_orig - Y_motion.
My questions are:
1. Is described procedure correct?
2. What should I use for my design matrix? There is a raw design
matrix: SPM.xX.X, but there is also filtered design in SPM.xX.xKXs.X
In addition, there is some scaling factor: SPM.xGX.gSF(1) (see
relevant post here:
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0702&L=SPM&P=R48764)
3. All my Y_motion values are negative, so I actually need to make
Y_cleaned = Y_orig + Y_motion
Thanks for help,
Vadim
|