where is the z or t statistic map?

AFNI Message Board
1

Dear AFNI experts,

I am new to AFNI and struggle to figure out where the statistic map is saved after running afni_proc.py without multiple testing. Is it one of the images within stats.${subject}_REML? I run afni_proc.py with the following options (canonical HRF with derivative):

afni_proc.py
-subj_id $subject
-script proc.$subject -scr_overwrite
-tcat_remove_first_trs 0
-dsets ${path_data}/${subject}_${task}bold.nii
-blur_size $smoothing.0
-regress_stim_types AM1
-regress_stim_times
${path_manage}/experimental_designs/${study}${exper_design}.txt
-regress_stim_labels boxcar_1
-regress_basis ‘SPMG2’
-regress_opts_3dD
-jobs 1
-regress_make_ideal_sum sum_ideal.1D
-volreg_align_to third
-regress_run_clustsim no
-regress_est_blur_epits
-regress_est_blur_errts
-regress_reml_exec
-regress_opts_reml -Rwherr whitened_errts.${subject}_REML

I guess this command of mine (adapted from some template) could be conciser…

Sorry for bothering you!

Best,

Wiktor

2

In the search for a t/z statistic map, I was looking for an image with high values in the visual cortex, as the experiment was with a visual stimulus: boxcar 20s off + 20s on. After checking today ‘sum_ideal.1D’ I realised that 3dDeconvolve did not convolve SPMG2 with the stimulus duration times, and that is why the fits were poor and nothing in the ‘stats.{subject}_REML+orig' output was looking like a visual cortex. After changing SPMG2 to SPMG2(20), the fits were as expected. However, I am still not sure where is the "ultimate" statistic map. Is it the first image (with index 0) in 'stats.{subject}_REML+orig’? Is it then an F-statistic map? Are then the other 8 images t-statistic maps for SPMG2 (2 images) and motion covariates (6 images)? Are the dof saved somewhere? In the attachment you can find the design plot generated by AFNI. Below there is my current command:

afni_proc.py
-subj_id $subject
-script proc.subject -scr_overwrite \ -tcat_remove_first_trs 0 \ -dsets {path_data}/{subject}_{task}bold.nii
-volreg_align_to third
-blur_size smoothing.0 \ -regress_stim_types AM1 \ -regress_stim_times \ {path_manage}/experimental_designs/AFNI{study}_{exper_design}.txt
-regress_stim_labels activation_stimulus
-regress_basis ‘SPMG2(20)’
-regress_opts_3dD
-jobs 1
-gltsym ‘SYM: activation_stimulus’
-glt_label 1 activity
-regress_make_ideal_sum sum_ideal.1D
-regress_run_clustsim no
-regress_est_blur_epits
-regress_est_blur_errts
-regress_reml_exec
-regress_opts_reml -Rwherr whitened_errts.${subject}_REML

Best,

Wiktor

X.jpg
X.jpg1536×2048 170 KB

3

Well, now I know that ‘3dinfo’ provides more information about the ‘stats.${subject}_REML+orig’ images/bricks than its AFNI header. Thus, I see that some bricks are coefficient maps rather than statistic maps. Yesterday I was thinking that all 9 bricks in my ‘stats.${subject}_REML+orig’ are F- and t- statistic maps, some of which referred to the motion covariates… Now I know that there are no images/bricks referring to the motion covariates.

I am only interested in extracting the statistic map corresponding to the HRF amplitude. Yesterday I called this statistic map the “ultimate” one. I am still not sure whether its the statistic map in brick [2]:

– At sub-brick #0 ‘activation_stimulus#0_Tstat’ datum type is float: -6.67307 to 22.7179
statcode = fitt; statpar = 88

or the one in brick [7]:

– At sub-brick #0 ‘activity#0_Tstat’ datum type is float: -8.30011 to 19.8636
statcode = fitt; statpar = 88

They are very similar to each other. Now I also know where the degrees of freedom are saved and how to convert a t-statistic map to a z-statistic map in AFNI:

https://afni.nimh.nih.gov/afni/community/board/read.php?1,67177,67179#msg-67179

I am not sure what is the difference between bricks [2] and [7] in my ‘stats.${subject}_REML+orig’. Could you please give a hint?

Best wishes,

Wiktor Olszowy