Choosing the right basis function (old data mystery)

Hello,

I have been given some data and within the afni analysis folder there are this file outlining the onset and duration of a stimulus presented in the first 6 runs (there are 9 runs in total):

45:30 103:33 166:36 238:39 309:33 382:39
45:30 103:33 166:36 238:39 309:33 382:39
38:30 96:33 159:36 231:39 302:33 375:39
45:30 103:33 166:36 238:39 309:33 382:39
45:30 103:33 166:36 238:39 309:33 382:39
45:30 103:33 166:36 238:39 309:33 382:39

The afni_proc.py (see below) for the data used for the task-regression_block has used -regress stim_types AM1 and dmUBLOCK(0) as the basis function. As far as I can tell, there is no reason for amplitude modulation in this experiment- should another basis function be used instead?


afni_proc.py -subj_id ${subj}         		                \
      -dsets ${subj}/afni_folder/f*${subj}_task*.nii.gz \
      -blocks despike tshift align tlrc volreg blur mask scale regress   \
      -copy_anat ${subj}/afni_folder/a*${subj}_T1w.nii.gz 	      \
      -align_opts_aea -cost lpc+ZZ                                \
      -tlrc_base MNI152_T1_2009c+tlrc                             \
      -tlrc_NL_warp                                               \
      -volreg_align_to MIN_OUTLIER                                \
      -volreg_align_e2a                                           \
      -volreg_tlrc_warp                                           \
      -volreg_warp_dxyz 2                                       \
      -mask_epi_anat yes                                          \
      -blur_in_automask                                           \
      -blur_to_fwhm                                               \
      -blur_size 4                                                \
      -regress_stim_times ${subj}/afni_folder/B*afni_task*.txt    \
      -regress_stim_types AM1		                                  \
                          AM1 AM1	                                \
                          AM1 AM1	                                \
                          AM1 AM1                                 \
                          AM1 AM1	                                \
                          AM1 AM1	                                \
      -regress_stim_labels                                        \
                      stim_on                                     \
                      rest15_on rest30_on                         \
                      stand15_on stand30_on                       \
                      step15_on  step30_on                        \
                      rest_sham  stand_sham                       \
                      step_sham  sham_on                          \
      -regress_local_times                                        \
      -regress_basis 'dmUBLOCK(0)'   			                        \
      -regress_apply_mot_types demean deriv                       \
      -regress_motion_per_run                                     \
      -regress_censor_motion 0.5                                 \
      -regress_compute_fitts                                      \
      -regress_reml_exec                                          \
      -regress_opts_3dD               		                        \
      -jobs 6					                                            \
      -GOFORIT 4                                                  \
      -num_glt 19                                                 \
              -gltsym 'SYM: +rest30_on -rest15_on'                         \
              -glt_label 1  rest30_vs_rest15                               \
              -gltsym 'SYM: +stand30_on -stand15_on'                       \
              -glt_label 2  stand30_vs_stand15                             \
              -gltsym 'SYM: +step30_on -step15_on'                         \
              -glt_label 3  step30_vs_step15                               \
              -gltsym 'SYM: +stand15_on -rest15_on'   		                 \
              -glt_label 4  stand_vs_rest15                   	           \
              -gltsym 'SYM: +step15_on -rest15_on'   		                   \
              -glt_label 5  step_vs_rest15                    	           \
              -gltsym 'SYM: +step15_on -stand15_on'   		                 \
              -glt_label 6  step_vs_stand15                    	           \
              -gltsym 'SYM: +stand30_on -rest30_on'   		                 \
              -glt_label 7  stand_vs_rest30                   	           \
              -gltsym 'SYM: +step30_on -rest30_on'   		                   \
              -glt_label 8  step_vs_rest30                    	           \
              -gltsym 'SYM: +step30_on -stand30_on'   		                 \
              -glt_label 9  step_vs_stand30                    	           \
              -gltsym 'SYM: +step30_on -stand30_on'   		                 \
              -glt_label 10  step_vs_stand30                    	         \
              -gltsym 'SYM: +stand30_on +stand15_on -rest15_on -rest30_on' \
              -glt_label 11  stand_vs_rest_overall                   	     \
              -gltsym 'SYM: +step30_on +step15_on -rest15_on -rest30_on'   \
              -glt_label 12  step_vs_rest_overall                   	     \
              -gltsym 'SYM: +step30_on +step15_on -stand15_on -stand30_on' \
              -glt_label 13  step_vs_stand_overall                         \
              -gltsym 'SYM: +rest15_on -rest_sham'                         \
              -glt_label 14  rest15_vs_rest_sham                           \
              -gltsym 'SYM: +stand15_on -stand_sham'                       \
              -glt_label 15  stand15_vs_step_sham                          \
              -gltsym 'SYM: +step30_on -step_sham'                         \
              -glt_label 16  step30_vs_step_sham                           \
              -gltsym 'SYM: +stim_on -sham_on'                             \
              -glt_label 17  stim_vs_sham                                  \
              -gltsym 'SYM: +stand_sham -rest_sham'                        \
              -glt_label 18  stand_vs_restSham                             \
              -gltsym 'SYM: +step_sham -rest_sham'                         \
              -glt_label 19  step_vs_restSham                              \
      -regress_est_blur_epits                                              \
      -regress_est_blur_errts                                              \
      -regress_run_clustsim yes

Cheers,
Gerome

Hi Gerome,

This looks good. The AM there is really to say there is some sort of modulator in the timing file. In this case, it is using duration modulation (which is why AM1 is used, rather than AM2, which would be more typical for amplitude modulation). Applying dmUBLOCK(0) (or just dmUBLOCK) will be the same as using dmBLOCK, except that the peak of a full (15+ second) convolution will be at 1, meaning the resulting beta weight would be the actual percent signal changed for a fully convolved event. That is also appropriate here, since the events are all longer than 15 seconds. If you plot the regressors, they should all scale up to 1.0.

This looks good to me.

  • rick

Thank you!