¥ Registration of each imaging run (there are 10): 3dvolreg
¥ Smooth each volume in space (136 sub-bricks per run): 3dmerge
¥ Create a brain mask: 3dAutomask and 3dcalc
¥ Rescale each voxel time series in each imaging run so that
its average through time is
100: 3dTstat and 3dcalc
H If baseline is 100, then a bq of 5 (say)
indicates a 5% signal change in that voxel at
time laq #q after stimulus
¥ Catenate all imaging runs together into one big dataset (1360
time points): 3dTcat
¥ Compute HRFs and statistics: 3dDeconvolve
H Each HRF will have 15 output points
(lags from 0 to 14) with a TR of 1.0 s, since the
input data has a TR of 2.0 s and we will be using the -stim_nptr k
r option with r=2
¥ Average together central points 4..9 of each separate HRF to get peak % change in each
voxel: 3dTstat