gdevenyi
commented
1 year ago From that pipeline, the second-level resampled DBM outputs give one image for each session/time point that encode the absolute or relative deformation to the subject-specific template.
Correct.
Because they are log-Jacobians, subtraction maps wouldn't be appropriate
I don't know what a subtraction map is.
but putting those into a statistical engine (SPM, FSL's randomise, PALM, etc) using a repeated measures design matrix would enable modeling the within subject slopes as well as group and time point differences (assuming a group x time experiment).
Our typical modelling is linear mixed-effect models within-subject (aka repeat measures) in RMINC.
Hi @gdevenyi ,
I have a project I'm consulting on for which the two-level DBM approach is worthwhile. I just want to clarify the outputs.
From that pipeline, the second-level resampled DBM outputs give one image for each session/time point that encode the absolute or relative deformation to the subject-specific template. Because they are log-Jacobians, subtraction maps wouldn't be appropriate to get change from timepoint to timepoint but putting those into a statistical engine (SPM, FSL's randomise, PALM, etc) using a repeated measures design matrix would enable modeling the within subject slopes as well as group and time point differences (assuming a group x time experiment).
Is that a correct understanding of the outputs that are produced?