How to use 3dMSS to analyze task-fMRI data?

AFNI Message Board
,
4

Roger,

Could you provide a more detailed analysis protocol for task-fMRI data? Meanwhile, what is the difference between 3dDeconvolve and " afni_proc.py -regree block"? It seems that both of them could fit the 4D data with ideal HRF. How could I combine the preprocess (tshift align tlrc volreg blur mask scale) with 3dDeconvolve and 3dMSS?

To clarify, 3dMSS functions as a program that infers the population-level hemodynamic response. The following emphasis aims to clarifies the confusions:

  • The methodology in 3dMSS aligns with conventional one- and two-sample t-tests in fMRI experiments, with a crucial distinction. Unlike t-tests, which assume the hemodynamic response is represented by its magnitude (or a scalar relative to the canonical HRF) through the regression coefficient (\beta) at the individual level, 3dMSS requires the expression of the hemodynamic response through its overall shape, which is derived from each individual. This shape can be estimated using basis functions such as TENT or CSPLIN through 3dDeconvolve at the individual level.

  • To estimate the hemodynamic response individually, you can either directly specify the basis function using 3dDeconvolve with options like -stim_times ... TENT..., or convey the information through the processing script afni_proc.py ... -regress_basis TENT....

These points are also illustrated in this blog post. Please let us know if this explanation brings clarity.

Gang Chen