Print

Print

I am currently using the longitudinal registration tool from SPM 12 in order to estimate GM volumes changes between 2 time points and relate these changes to clinical measures. 

Once I created the Jacobian difference image, I understood I should multiply the volume differences image by the GM volumes but I can't get how. 

1/ Which GM volume shall I choose (Time 1 or the mid point average) ?

To avoid bias, the longitudinal registration aligns the scans to the mid-point average. All the Jacobians etc are in this average space.  Therefore, I'd suggest that for any segmenting, you should use the mid-point average.

2/ Shall I use the total GM volume or voxel by voxel ?

The aim is to identify regional grey matter volume changes, so you'd multiply the GM maps by the volume change maps (Jacobian differences) to give maps of GM volume change.

3/ Which tool shall I use to do this in SPM ? 

If you don't have so many subjects, you could multiply the images together via ImCalc (evaluating i1.*i2).  If there are many subjects, you may want to try some form of MATLAB scripting. 

Best regards,
-John