Dear AFNI minds,
I am interested in using 3dDeconvolve to dump out trial-type-averaged hemodynamic responses in a VOI from my event-related paradigm.
Each trial has four events/stimuli in it, where these events are separated by a jittered, variable interval (2-6 s). Including the ITI between trials.
The scan has a 1 s TR.
My 3dDeconvolve script is currently set to obtain seven volumes (0-6), as indicated in the excerpt below. The modeling is applied to a 3d+time dataset normalized as percent signal change.
This maxlag of 6 (TR or s) is to essentially to capture an estimate of peak signal change time-locked to each event type out to 6 seconds, WITHOUT much interference from the subsequent event of the trial, which occurs a mean 4 s later.
-stim_file 1 ../../../vectors/'cue_neuTR.1D' -stim_maxlag 1 6 -stim_label 1 'ant_neu' \
-stim_file 2 ../../../vectors/'cue_lorewTR.1D' -stim_maxlag 2 6 -stim_label 2 'ant_lorew' \
-stim_file 3 ../../../vectors/'cue_hirewTR.1D' -stim_maxlag 3 6 -stim_label 3 'ant_hirew' \
-stim_file 4 ../../../vectors/'cue_lolossTR.1D' -stim_maxlag 4 6 -stim_label 4 'ant_loloss' \
-stim_file 5 ../../../vectors/'cue_hilossTR.1D' -stim_maxlag 5 6 -stim_label 5 'ant_hiloss' \
-stim_file 6 ../../../vectors/'targ_neuTR.1D' -stim_maxlag 6 6 -stim_label 6 'targ_neu' \
-stim_file 7 ../../../vectors/'targ_lorewTR.1D' -stim_maxlag 7 6 -stim_label 7 'targ_lorew' \
-stim_file 8 ../../../vectors/'targ_hirewTR.1D' -stim_maxlag 8 6 -stim_label 8 'targ_hirew' \
-stim_file 9 ../../../vectors/'targ_lolossTR.1D' -stim_maxlag 9 6 -stim_label 9 'targ_loloss' \
-stim_file 10 ../../../vectors/'targ_hilossTR.1D' -stim_maxlag 10 6 -stim_label 10 'targ_hiloss' \
-stim_file 11 ../'fbkTRmatrix.1D[0]' -stim_maxlag 11 6 -stim_label 11 'fbk_neuhit' \
-stim_file 12 ../'fbkTRmatrix.1D[1]' -stim_maxlag 12 6 -stim_label 12 'fbk_neumiss' \
-stim_file 13 ../'fbkTRmatrix.1D[2]' -stim_maxlag 13 6 -stim_label 13 'fbklrhit' \
-stim_file 14 ../'fbkTRmatrix.1D[3]' -stim_maxlag 14 6 -stim_label 14 'fbklrmiss' \
-stim_file 15 ../'fbkTRmatrix.1D[4]' -stim_maxlag 15 6 -stim_label 15 'fbkhrhit' \
-stim_file 16 ../'fbkTRmatrix.1D[5]' -stim_maxlag 16 6 -stim_label 16 'fbkhtmiss' \
-stim_file 17 ../'fbkTRmatrix.1D[6]' -stim_maxlag 17 6 -stim_label 17 'fbklphit' \
-stim_file 18 ../'fbkTRmatrix.1D[7]' -stim_maxlag 18 6 -stim_label 18 'fbklpmiss' \
-stim_file 19 ../'fbkTRmatrix.1D[8]' -stim_maxlag 19 6 -stim_label 19 'fbkhphit' \
-stim_file 20 ../'fbkTRmatrix.1D[9]' -stim_maxlag 20 6 -stim_label 20 'fbkhipmiss' \
-stim_file 21 ../'3dmotion.1D[1]' -stim_label 21 'roll' \
-stim_file 22 ../'3dmotion.1D[2]' -stim_label 22 'pitch' \
-stim_file 23 ../'3dmotion.1D[3]' -stim_label 23 'yaw' \
-stim_file 24 ../'3dmotion.1D[4]' -stim_label 24 'dS' \
-stim_file 25 ../'3dmotion.1D[5]' -stim_label 25 'dL' \
-stim_file 26 ../'3dmotion.1D[6]' -stim_label 26 'dP' \
-tout -bucket 6HRFmaps \
-iresp 1 ./6antneu \
-iresp 2 ./6antlorew \
-iresp 3 ./6anthirew ....
Am I correct in assuming that if I went with stim_maxlag 10 or 14 to potentially dump out a full-length HRF for each event type, I would be contaminating these -iresp output estimates with effects of the subsequent event, even though I jitter?
So, my question is, is whether the shorter I specify the maxlag, the more accurate a picture I'll get of true signal change in those initial seconds after an event type?
It would be nice to be able to let the brain relax for a full 14 s HRF between each stimulus within a trial, but a mean 56 s trial length is not an option, so I'm jittering to try to tease out trial subcomponent responses.
Jim