AFNI version info (afni -ver
): AFNI_23.0.07
Hello,
We are using pre-treatment brain activations to predict symptom change over time during treatment. Our 3dLMEr script is pasted below, with two subjects' (deidentifed) data shown as an example. The variable InputFile points to "blocks" of symptom data that we created for this purpose; for example, if subject 101 at timepoint 1 had a GAD-7 score of 5, the file GAD7/101-GAD7-Wk1+tlrc would have a value of 5 at every voxel. The variable VoxelActivity_T1 points to our pre-treatment fMRI data at the sub-brik of interest.
The 3dLMEr script ran successfully, and we are now wondering what ACF parameters to use in 3dClustSim. We ran 3dFWHMx on the residuals from 3dLMEr, then put this into 3dClustSim, and got a very low minimum cluster size. We then ran 3dFWHMx on each individual stats file and put the averaged parameters into 3dClustSim instead, which gave us a somewhat more reasonable minimum cluster size, but we are still looking at a lot more significant clusters than we expected. We have seen this across multiple different fMRI tasks, and it is making us wonder if we should be thinking differently about significance thresholding in this type of analysis. Since we are predicting "perfectly smooth" data, e.g., the block of 5's, does this mean we should choose ACF parameters in a different way? Or have we gone wrong somewhere else?
Thank you,
Hannah
3dLMEr -prefix _MID_WB-GAD7_a_LME_Num \
-jobs 48 \
-model 'Time*Treatment*VoxelActivity_T1+enorm+GAD7_T1+(1|Subj)' \
-resid errts._MID_WB-GAD7_a_LME_Num \
-qVars 'enorm,Time,GAD7_T1' \
-qVarsCenters '0.1,5.99,11.97' \
-vVars 'VoxelActivity_T1' \
-SS_type 3 \
-mask MNI_mask_resampled+tlrc \
-dataTable \
Subj enorm Time Treatment VoxelActivity_T1 GAD7_T1 InputFile \
101 0.1298 1 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk1+tlrc \
101 0.1298 2 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk2+tlrc \
101 0.1298 3 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk3+tlrc \
101 0.1298 4 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk4+tlrc \
101 0.1298 5 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk5+tlrc \
101 0.1298 6 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk6+tlrc \
101 0.1298 7 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk7+tlrc \
101 0.1298 8 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk8+tlrc \
101 0.1298 9 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk9+tlrc \
101 0.1298 10 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-Wk10+tlrc \
101 0.1298 11 BA ../../stats.101-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 10 GAD7/101-GAD7-PostWDLOCF+tlrc \
102 0.0643 1 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk1+tlrc \
102 0.0643 2 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk2+tlrc \
102 0.0643 3 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk3+tlrc \
102 0.0643 4 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk4+tlrc \
102 0.0643 5 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk5+tlrc \
102 0.0643 6 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk6+tlrc \
102 0.0643 7 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk7+tlrc \
102 0.0643 8 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk8+tlrc \
102 0.0643 9 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk9+tlrc \
102 0.0643 10 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-Wk10+tlrc \
102 0.0643 11 BA ../../stats.102-T1-_MID+tlrc'[a.rall_GLT#0_Coef]' 8 GAD7/102-GAD7-PostWDLOCF+tlrc \