IŽd like to compare two timepoints with fslvbm, therefore I created a "halfway-space" average GM image of each subject using siena_flirt/pairreg, created a study-specific template, and then fnirted the "halfway-space" to the template. Now I wonder whether I should for every timepoint
- apply in one step the rigid-body-transformation of timepoint_to_halfway + the nonlinear warp-halfway_to_template (e.g. with applywarp --premat -w)
- then multiply the output with the Jacobian of fnirt-halfway_to_template
- first apply the rigid-body transformation, then on this output the nonlinear warp, and then multiply with the Jacobian?
I tried both and got the same GM "shape", but slightly different intensity values. Do you have an advice, is the one-step-transformation better for some reason? And, I don`t need to add a "scaling factor" for the first rigid-body-transformation, correct?
Many thanks for your help!