Hello,
I am trying to do a univariate contrast with three runs of data, but the TR length of my first run (2.25 s) differs from the TR length of the second and third runs (2.8 s). Is it possible to do this with 3ddeconvolve? I just ran my analysis as follows:
3dDeconvolve -input functional_loc_ns+orig.HEAD functional_enc_r02_ns+orig.HEAD functional_enc_r03_ns+orig.HEAD -polort A \
-num_stimts 9 -basis_normall 1 -float -local_times \
-stim_times 1 ./NoCond/timing_DutchWord.1D 'BLOCK(2.6,1)' \
-stim_label 1 enc_DutchWord \
-stim_times 2 ./Word_Nonword/time_nonWord.1D 'BLOCK(4.5,1)' \
-stim_label 2 loc_Nonword\
-stim_times 3 ./Word_Nonword/time_word.1D 'BLOCK(4.5,1)' \
-stim_label 3 loc_Word\
-stim_file 4 'dfile.rall.1D[0]' -stim_base 4 -stim_label 4 roll \
-stim_file 5 'dfile.rall.1D[1]' -stim_base 5 -stim_label 5 pitch \
-stim_file 6 'dfile.rall.1D[2]' -stim_base 6 -stim_label 6 yaw \
-stim_file 7 'dfile.rall.1D[3]' -stim_base 7 -stim_label 7 dS \
-stim_file 8 'dfile.rall.1D[4]' -stim_base 8 -stim_label 8 dL \
-stim_file 9 'dfile.rall.1D[5]' -stim_base 9 -stim_label 9 dP \
-gltsym 'SYM: enc_DutchWord-loc_Nonword' \
-glt_label 1 DutchMinNonword \
-tout -fout -bucket decon_Dutch_Nonword
and received the following output:
++ 3dDeconvolve: AFNI version=AFNI_19.2.10 (Aug 7 2019) [64-bit]
++ Authored by: B. Douglas Ward, et al.
++ loading dataset functional_loc_ns+orig.HEAD functional_enc_r02_ns+orig.HEAD functional_enc_r03_ns+orig.HEAD
++ Auto-catenated input datasets treated as multiple imaging runs
++ Auto-catenated datasets start at: 0 201 370
++ STAT automask has 41137 voxels (out of 316800 = 13.0%)
++ Skipping check for initial transients
++ Imaging duration=452.2 s; Automatic polort=4
++ -stim_times using TR=2.25 s for stimulus timing conversion
++ -stim_times using TR=2.25 s for any -iresp output datasets
++ [you can alter the -iresp TR via the -TR_times option]
++ -stim_times 1 using LOCAL times
*+ WARNING: '-stim_times 1' (LOCAL) run#2 has 13 times outside range 0 .. 378 [PSFB syndrome]
+ dataset TR being used is 2.25 s -- unusable times follow
383.21 388.827 391.686 397.303 400.144 408.553 414.17 428.179 431.038 439.446 442.288 445.147 448.005
*+ WARNING: '-stim_times 1' (LOCAL) run#3 has 16 times outside range 0 .. 378 [PSFB syndrome]
+ dataset TR being used is 2.25 s -- unusable times follow
380.154 385.771 391.388 402.605 405.464 411.081 413.923 416.765 419.623 425.24 436.457 439.299 442.141 444.983 450.6 453.459
++ -stim_times 2 using LOCAL times
++ -stim_times 3 using LOCAL times
++ INFO: -gltsym: isolated '-' is merged with following 'loc_Nonword'
------------------------------------------------------------
GLT matrix from 'SYM: enc_DutchWord - loc_Nonword':
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0
++ Number of time points: 539 (no censoring)
+ Number of parameters: 24 [21 baseline ; 3 signal]
++ Memory required for output bricks = 16,473,600 bytes (about 16 million)
++ Wrote matrix values to file decon_Dutch_Nonword.xmat.1D
++ ========= Things you can do with the matrix file =========
++ (a) Linear regression with ARMA(1,1) modeling of serial correlation:
3dREMLfit -matrix decon_Dutch_Nonword.xmat.1D \
-input "functional_loc_ns+orig.HEAD functional_enc_r02_ns+orig.HEAD functional_enc_r03_ns+orig.HEAD" \
-fout -tout -Rbuck decon_Dutch_Nonword_REML -Rvar decon_Dutch_Nonword_REMLvar -verb
++ N.B.: 3dREMLfit command above written to file decon_Dutch_Nonword.REML_cmd
++ (b) Visualization/analysis of the matrix via ExamineXmat.R
++ (c) Synthesis of sub-model datasets using 3dSynthesize
++ ==========================================================
++ ----- Signal+Baseline matrix condition [X] (539x24): 5.19594 ++ VERY GOOD ++
++ ----- Signal-only matrix condition [X] (539x3): 1.11579 ++ VERY GOOD ++
++ ----- Baseline-only matrix condition [X] (539x21): 5.16435 ++ VERY GOOD ++
++ ----- stim_base-only matrix condition [X] (539x6): 3.48693 ++ VERY GOOD ++
++ ----- polort-only matrix condition [X] (539x15): 1.01973 ++ VERY GOOD ++
++ +++++ Matrix inverse average error = 1.9655e-15 ++ VERY GOOD ++
++ Matrix setup time = 0.48 s
++ Calculations starting; elapsed time=9.647
++ voxel loop:0123456789.0123456789.0123456789.0123456789.0123456789.
++ Calculations finished; elapsed time=47.726
++ Smallest FDR q [0 Full_Fstat] = 2.55031e-12
++ Smallest FDR q [2 enc_DutchWord#0_Tstat] = 0.0743257
++ Smallest FDR q [3 enc_DutchWord_Fstat] = 0.0743186
++ Smallest FDR q [5 loc_Nonword#0_Tstat] = 2.02278e-12
++ Smallest FDR q [6 loc_Nonword_Fstat] = 2.02267e-12
++ Smallest FDR q [8 loc_Word#0_Tstat] = 1.13607e-11
++ Smallest FDR q [9 loc_Word_Fstat] = 1.13587e-11
++ Smallest FDR q [11 DutchMinNonword_GLT#0_Tstat] = 4.54299e-12
++ Smallest FDR q [12 DutchMinNonword_GLT_Fstat] = 4.54299e-12
++ Wrote bucket dataset into ./decon_Dutch_Nonword+orig.BRIK
+ created 9 FDR curves in bucket header
++ #Flops=1.84974e+11 Average Dot Product=42.282
It seems that the TR length is pulled from the HEAD file of the first input file (2.25 s) and applying that length to all input files. Do you have any recommendations?
Thank you!
Heather
