fMRI Fear Conditioning Model Overview - PennBBL/conte GitHub Wiki

Motivation:

To detail the model development for analysis of the fMRI Fear Conditioning paradigm (version3/design3 only). Analyses were run on run1 only by MQ but scripts will be analogous for run2 reversal if someone were to set them up. The fMRI fear conditioning paradigm is detailed in the wiki "fMRI Fear Conditioning Paradigm" and the scripts to process these models and achieve the output are detailed in the wiki "fMRI Fear Conditioning FSl Analysis Pipeline".

fMRI Model Overview:

Various fMRI models were run with the main contrast of interest being activation to aversive face when scream is absent relative to neutral face (CS+u > CS-) or the inverse (CS- > CS+u), as is consistent with most fear conditioning literature.

The first-level feat model types focused on for run1 fear conditioning were:

  • Mini Block
  • Linear model
  • Split Half Cue and Outcome
  • PPI (for 3 ROIs of interest)

For group-level (higher-level) feat models we focused on:

  • Single Group Average (averaging across all subjects not taking into account diagnostic group)
  • Diagnostic Group Difference (various combinations of the three diagnostic groups)
  • Symptom Correlations (both controlling for diagnostic group and not controlling for diagnostic group)

Mini Block:

The mini block model consists of 12 first-level or subject-level contrasts:

  • cope1 Aversive face with tone (CS+p)
  • cope2 Aversive face without tone (CS+u)
  • cope3 Neutral face (CS-)
  • cope4 Aversive face with tone (CS+p) linear time modeling
  • cope5 Aversive face without tone linear (CS+u) time modeling
  • cope6 Neutral face (CS-) with linear time modeling
  • cope7 Any Stimulus (CS+p, CS+u, CS-)
  • cope8 Any Stimulus (CS+p, CS+u, CS-) with linear time modeling
  • cope9 Catch trials
  • cope10 Flashing screen
  • cope11 Aversive face without tone (CS+u) > Neutral face (CS-)
  • cope12 Aversive face without tone (CS+u) with linear time modeling > Neutral face (CS-) with linear time modeling

The stimulus onset (stick) files used are:

  • CS+p
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_mini_block.txt
  • CS+u
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_mini_block.txt
  • CS-
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_mini_block.txt
  • CS+p linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_mini_block_linear.txt
  • CS+u linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_mini_block_linear.txt
  • CS- linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_mini_block_linear.txt
  • Catch trials
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/catch.txt
  • Flashing screen
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/flashing_screen.txt

Note that there is a different directory for each fMRI fear conditioning order (0_2 and 1_3) and this will determine which face is aversive or neutral (face1 or face2). The easy way to tell is that if for example there is a face1_aversive_mini_block.txt file, then that means that face1_notone_mini_block.txt is CS+u and face2_notone_mini_block.txt is neutral (CS-).

The timing is set up so that each of these stimuli is modeled for 6 seconds or the full duration that the face is present. The flashing screen is also modeled for 6 seconds. (Note: The aversive tone is still modeled for one second as it is only present for 1 second, and catch trials are modeled for 2 seconds).

The linear time modeling (for stimuli like cope4) is achieved by a symmetrical time shift in the third column (should word this better by DW). For example, this stimulus below is the CS+u without linear modeling (cope2), notice how the third column is all 1's, denoting a constant.

    40	6	1  
    52	6	1  
    152	6	1  
    178	6	1  
    196	6	1  
    212	6	1  
    220	6	1  
    284	6	1  
    354	6	1  
    372	6	1  
    396	6	1  
    472	6	1

Now here is the same stimulus (CS+u) but with linear time modeling (cope5). Notice how the third column now goes from -5.5 to 5.5 with consistent steps of 1, centering around the stimulus onset in the middle.

    40	6	-5.5  
    52	6	-4.5  
    152	6	-3.5  
    178	6	-2.5  
    196	6	-1.5  
    212	6	-0.5  
    220	6	0.5  
    284	6	1.5  
    354	6	2.5  
    372	6	3.5  
    396	6	4.5  
    472	6	5.5

You'll notice on monstrum some stick files and feats with the label "linear_scaled" this is because we realized that the ---------- were unbalanced because one appears 12 times and one appears 24 times, so in an attempt to fix that we .... But this didn't seem to make much of a difference and so we stuck with the original linear modeling. DW wording

On CFN these first-level model outputs are labeled as:

xxxxx

Note: on monstrum they have the naming convention of: [scanid]_[run1 nifti name]_bbr_linear_mini_block.feat
For example: /import/monstrum/conte_815814/subjects/87601_9952/10_bbl1_fearcondV3R1_2.5x2.5x2.5_176/nifti/009952_bbl1_fearcondV3R1_2.5x2.5x2.5_176_SEQ010_bbr_linear_mini_block.feat

Linear Model:

The linear time model consists of 24 first-level or subject-level contrasts:

  • cope1 Aversive face with tone (CS+p) cue (first second)
  • cope2 Aversive scream (fifth second of CS+p)
  • cope3 Aversive face without tone (CS+u) cue (first second)
  • cope4 Aversive face without tone (CS+u) outcome (fifth second)
  • cope5 Neutral face (CS-) cue (first second)
  • cope6 Neutral face (CS-) outcome (fifth second)
  • cope7 Aversive face with tone (CS+p) cue (first second) linear time modeling
  • cope8 Aversive scream (fifth second of CS+p) linear time modeling
  • cope9 Aversive face without tone (CS+u) cue (first second) linear time modeling
  • cope10 Aversive face without tone (CS+u) outcome (fifth second) linear time modeling
  • cope11 Neutral face (CS-) cue (first second) linear time modeling
  • cope12 Neutral face (CS-) outcome (fifth second) linear time modeling
  • cope13 Catch Trials
  • cope14 Flashing Screen
  • cope14 Face > Aversive Scream
  • cope16 Any Stimulus (CS+p, CS+u, CS-) cue (first second)
  • cope17 Any Stimulus (CS+p, CS+u, CS-) outcome (fifth second)
  • cope18 Any Stimulus (CS+p, CS+u, CS-) outcome (fifth second) > Any Stimulus (CS+p, CS+u, CS-) cue (first second)
  • cope19 Aversive face without tone (CS+u) cue (first second) > Neutral face (CS-) cue (first second)
  • cope20 Aversive face without tone (CS+u) outcome (fifth second) > Neutral face (CS-) outcome (fifth second)
  • cope21 Aversive face without tone (CS+u) cue (first second) linear time modeling > Neutral face (CS-) cue (first second) linear time modeling
  • cope22 Aversive face without tone (CS+u) outcome (fifth second) linear time modeling > Neutral face (CS-) outcome (fifth second) linear time modeling
  • cope23 Aversive face without tone (CS+u) > Neutral face (CS-)
  • cope24 Aversive face without tone (CS+u) linear time modeling > Neutral face (CS-) linear time modeling

The stimulus onset (stick) files used are:

  • CS+p cue
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec1.txt
  • CS+p outcome
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec5.txt
  • CS+u cue
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1.txt
  • CS+u outcome
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5.txt
  • CS- cue
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1.txt
  • CS- outcome
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5.txt
  • CS+p cue linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec1_linear.txt
  • CS+p outcome linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec5_linear.txt
  • CS+u cue linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_linear.txt
  • CS+u outcome linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_linear.txt
  • CS- cue linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_linear.txt
  • CS- outcome linear
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_linear.txt
  • Catch trials
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/catch.txt
  • Flashing screen
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/flashing_screen.txt
  • Any Stimulus cue
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/anyface_firstsec.txt
  • Any Stimulus outcome
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/anyface_lastsec.txt

Note that there is a different directory for each fMRI fear conditioning order (0_2 and 1_3) and this will determine which face is aversive or neutral (face1 or face2). The easy way to tell is that if for example there is a face1_aversive_mini_block.txt file, then that means that face1_notone_linear.txt is CS+u and face2_notone_linear.txt is neutral (CS-).

The timing is set up so that each of these stimuli is modeled for half a second to mimic an instantaneous stimulus response. The flashing screen is modeled for 6 seconds, the catch trials are modeled for 2 seconds. The aversive tone is still modeled for one second. Note: the linear modeling for those stimuli modeled like that (i.e. cope7) is achieved the same way as described for the mini block model above.

On CFN these first-level model outputs are labeled as:

xxxxx

Note: on monstrum they have the naming convention of: [scanid]_[run1 nifti name]_bbr_linear.feat
For example: /import/monstrum/conte_815814/subjects/87601_9952/10_bbl1_fearcondV3R1_2.5x2.5x2.5_176/nifti/009952_bbl1_fearcondV3R1_2.5x2.5x2.5_176_SEQ010_bbr_linear.feat

Split Half Cue and Outcome:

The split half cue and outcome model consists of 32 first-level or subject-level contrasts:

  • cope1 Aversive face with tone (CS+p) cue (first second) first half of the task
  • cope2 Aversive scream (fifth second of CS+p) first half of the task
  • cope3 Aversive face without tone (CS+u) cue (first second) first half of the task
  • cope4 Aversive face without tone (CS+u) outcome (fifth second) first half of the task
  • cope5 Neutral face (CS-) cue (first second) first half of the task
  • cope6 Neutral face (CS-) outcome (fifth second) first half of the task
  • cope7 Aversive face with tone (CS+p) cue (first second) second half of the task
  • cope8 Aversive scream (fifth second of CS+p) second half of the task
  • cope9 Aversive face without tone (CS+u) cue (first second) second half of the task
  • cope10 Aversive face without tone (CS+u) outcome (fifth second) second half of the task
  • cope11 Neutral face (CS-) cue (first second) second half of the task
  • cope12 Neutral face (CS-) outcome (fifth second) second half of the task
  • cope13 Aversive face with tone (CS+p) cue (first second) first half of the task > Aversive face without tone (CS+u) cue (first second) first half of the task
  • cope14 Aversive face with tone (CS+p) outcome (fifth second) first half of the task > Aversive face without tone (CS+u) outcome (fifth second) first half of the task
  • cope15 Aversive face with tone (CS+p) cue (first second) second half of the task > Aversive face without tone (CS+u) cue (first second) second half of the task
  • cope16 Aversive face with tone (CS+p) outcome (fifth second) second half of the task > Aversive face without tone (CS+u) outcome (fifth second) second half of the task
  • cope17 Aversive face with tone (CS+p) cue (first second) first half of the task > Neutral face (CS-) cue (first second) first half of the task
  • cope18 Aversive face with tone (CS+p) outcome (fifth second) first half of the task > Neutral face (CS-) outcome (fifth second) first half of the task
  • cope19 Aversive face with tone (CS+p) cue (first second) second half of the task > Neutral face (CS-) cue (first second) second half of the task
  • cope20 Aversive face with tone (CS+p) outcome (fifth second) second half of the task > Neutral face (CS-) outcome (fifth second) second half of the task
  • cope21 Aversive face without tone (CS+u) cue (first second) first half of the task > Neutral face (CS-) cue (first second) first half of the task
  • cope22 Aversive face without tone (CS+u) outcome (fifth second) first half of the task > Neutral face (CS-) outcome (fifth second) first half of the task
  • cope23 Aversive face without tone (CS+u) cue (first second) second half of the task > Neutral face (CS-) cue (first second) second half of the task
  • cope24 Aversive face without tone (CS+u) outcome (fifth second) second half of the task > Neutral face (CS-) outcome (fifth second) second half of the task
  • cope25 Aversive face without tone (CS+u) cue (first second) second half of the task > Aversive face without tone (CS+u) cue (first second) first half of the task
  • cope26 Aversive face without tone (CS+u) outcome (fifth second) second half of the task > Aversive face without tone (CS+u) outcome (fifth second) first half of the task
  • cope27 Catch Trials
  • cope28 Neutral face (CS-) cue (first second) second half of the task > Neutral face (CS-) cue (first second) first half of the task
  • cope29 Neutral face (CS-) outcome (fifth second) second half of the task > Neutral face (CS-) outcome (fifth second) first half of the task
  • cope30 (Aversive face without tone (CS+u) cue (first second) second half of the task > Aversive face without tone (CS+u) cue (first second) first half of the task) > (Neutral face (CS-) cue (first second) second half of the task > Neutral face (CS-) cue (first second) first half of the task)
  • cope31 (Aversive face without tone (CS+u) outcome (fifth second) second half of the task > Aversive face without tone (CS+u) outcome (fifth second) first half of the task) > (Neutral face (CS-) outcome (fifth second) second half of the task > Neutral face (CS-) outcome (fifth second) first half of the task)
  • cope32 Task Contrast

The stimulus onset (stick) files used are:

  • CS+p cue first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec1_first_half.txt
  • Aversive scream first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/aversive_tone_first_half.txt
  • CS+u cue first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_first_half.txt
  • CS+u outcome first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_first_half.txt
  • CS- cue first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_first_half.txt
  • CS- outcome first half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_first_half.txt
  • CS+p cue second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_aversive_sec1_second_half.txt
  • Aversive scream second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/aversive_tone_second_half.txt
  • CS+u cue second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_second_half.txt
  • CS+u outcome second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_second_half.txt
  • CS- cue second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec1_second_half.txt
  • CS- outcome second half
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/face[1/2]_notone_sec5_second_half.txt
  • Catch trials
    /data/joy/BBL/studies/conte/fmriDesignFiles/order[0_2/1_3]/run1/stick_files/catch.txt

Note that there is a different directory for each fMRI fear conditioning order (0_2 and 1_3) and this will determine which face is aversive or neutral (face1 or face2). The easy way to tell is that if for example there is a face1_aversive_mini_block.txt file, then that means that face1_notone_linear.txt is CS+u and face2_notone_linear.txt is neutral (CS-).

The timing is set up so that each of these stimuli is modeled for half a second to mimic an instantaneous stimulus response, but the original stimulus (stick) files are now split into first and second halves of the task. The flashing screen is modeled for 6 seconds, the catch trials are modeled for 2 seconds. The aversive tone is still modeled for one second. Note: the linear modeling for those stimuli modeled like that is achieved the same way as described for the mini block model above.

On CFN these first-level model outputs are labeled as:

xxxxx

Note: on monstrum they have the naming convention of: [scanid]_[run1 nifti name]_bbr_split_half_sec1_sec5_NEW_MODEL.feat
For example: /import/monstrum/conte_815814/subjects/87601_9952/10_bbl1_fearcondV3R1_2.5x2.5x2.5_176/nifti/009952_bbl1_fearcondV3R1_2.5x2.5x2.5_176_SEQ010_bbr_split_half_sec1_sec5_NEW_MODEL.feat

FIR Models:

Finite-impulse-response (FIR) models were run to examine in greater detail the timing signature of the brain during the task (particularly the three ROIs of interest). Please see the FSL Feat wiki for more information about FIR model setup in feat. These analyses were not run on CFN (due to time the fact that they were more exploratory and not used for more final analyses).

However, on monstrum, there were several varying FIR models run- all with phase shift of 0, and with 4, 5, or 6 timepoints. The 5 timepoint model was ultimately deemed to be the best fit for the data. On monstrum the subject-level output has the naming convention of: [scanid]_[run1 nifti name]_fir_phase0_5tpt.feat
For example: /import/monstrum/conte_815814/subjects/100936_10307/9_bbl1_fearcondV3R1_2.5x2.5x2.5_176/nifti/010307_bbl1_fearcondV3R1_2.5x2.5x2.5_176_SEQ09_fir_phase0_5tpt.feat

I won't go into too much detail here about this model but essentially it chunks the timeseries into 5 timepoints by setting the Convolution of each EV (stick file in first-level feat) to "FIR basis functions" then the phase to 0, the number to 5, and the window to 15. Check out the design.fsf in the example directory above for more detail on how the model was set up. These models were run using the pipeline described in the FSL Analysis Pipeline wiki but on monstrum (script: /import/monstrum/conte_815814/scripts/functional/design3_run_feat.sh)

Diagnostic Group Difference Higher-Level Contrasts:

The first-level models were run as group level models examining diagnostic group differences. These higher-level feats have the following contrasts for each first-level cope (i.e. 3 group mean for CS+u, 3 group mean for CS-, etc.)

  • cope1 3 group mean (averaged across all three groups)
  • cope2 NC>CR
  • cope3 NC>SZ
  • cope4 CR>SZ
  • cope5 NC>CR+SZ (combined clinical groups)
  • cope6 NC+CR>SZ (combined NC and CR groups)
  • cope7 NC only
  • cope8 CR only
  • cope9 SZ only

Clinical Symptom Higher-Level Modeling:

The first-level models were also run as group level models examining correlation of the task with clinical symptoms (CAINS, SIPS positive symptoms, and STAI anxiety) both with and without group regressed out. These higher-level feats have the following contrasts for each first-level cope (i.e. 3 group mean for CS+u, 3 group mean for CS-, etc.)

  • cope1 group mean
  • cope2 clinical symptom correlation (CAINS/STAI/SIPS depending on the model run)

The design matrix set up determines if diagnostic group is controlled for or not. If group is controlled for, the clinical symptom is demeaned within group. Otherwise, the symptom is demeaned across the whole sample.

For example, here is an excerpt from a group controlled design matrix where the first 3 columns describe group (NC, CR, SZ) then the next three columns are the CAINS score demeaned within group:

      /NumWaves	6  
      /NumPoints	96  
      /PPheights		1.000000e+00	1.000000e+00  
                  
      /Matrix  
          1.0000         0         0   -4.0882         0         0  
          1.0000         0         0   -5.0882         0         0  
               0         0    1.0000         0         0   -2.4737  
               0    1.0000         0         0   -2.9714         0  
          1.0000         0         0   -4.0882         0         0   
               0         0    1.0000         0         0    6.5263

Compare this to a model where group is not controlled for, notice the group columns (first 3 group columns) are the same but the CAINS column is a single column demeaned across the whole sample:

      /NumWaves	4  
      /NumPoints	96  
      /PPheights		1.000000e+00	1.000000e+00	1.000000e+00 1  
          
      /Matrix
          1.0000         0         0   -8.9828  
          1.0000         0         0   -9.9828  
               0         0    1.0000    6.0172  
               0    1.0000         0   -1.9828  
          1.0000         0         0   -8.9828

ROI Selection:

The three main ROI's used for analyses were the ventromedial prefrontal cortex (vmPFC), bilateral anterior insular cortex (AIC), and bilateral amygdala. ROI's were chosen on an a priori basis from the CONTE grant as well as based on prior fear conditioning literature (see "Fullana et al. 2016" for more details).

ROI's were created in several different ways however the final ROI segmentation was created using "neurosynth" in the following manner:

vmPFC:

The vmPFC ROI was created by searching the term "vmpfc" in neurosynth, selecting the vmpfc result which contains 143 studies, then downloading the resulting map with reverse inference. Once on the server, the ROI image was eroded once, and then dilated once until it appeared to match the anatomical vmPFC (decided by DW) and then binarized.

The final vmPFC ROI is saved here:

AIC:

The final AIC bilateral ROI is saved here:

Amygdala:

The bilateral amygdala ROI was created by searching the term "amygdala" in neurosynth, selecting the amygdala result which contains 1,245 studies, then downloading the resulting map with reverse inference. Once on the server, the ROI image was thresholded at 20 and above so it appeared to match the anatomical amygdala (decided by DW) and then binarized.

The final amygdala bilateral ROI is saved here:

PPI:

Psychophysiological interaction (PPI) models were run to examine in greater detail the connectivity of the three regions detailed above with the rest of the brain. Please see the FSL PPI wiki for more information about PPI model setup in feat. These analyses were not run on CFN (due to time the fact that they were more exploratory and not used for more final analyses).

However, on monstrum, there were several PPI models run- all with varying ROI parcellations of the amygdala, anterior insula, or vmPFC. On monstrum the subject-level outputs have the naming convention of: [scanid]_[run1 nifti name]bbr[roi]_mq.feat
For example: /import/monstrum/conte_815814/subjects/100936_10307/9_bbl1_fearcondV3R1_2.5x2.5x2.5_176/nifti/010307_bbl1_fearcondV3R1_2.5x2.5x2.5_176_SEQ09_bbr_vmpfc_neurosynth_20170517_ppi_mini_block.feat

I won't go into too much detail here about this model but essentially it takes the ROI timeseries for a subject (created by the script /import/monstrum/conte_815814/scripts/functional/PPI_model/extract_roits_mq.sh) and runs a feat (using the script /import/monstrum/conte_815814/scripts/functional/PPI_model/PPI_model_mq.sh) with the EVs as per usual (i.e. CS+u, CS+p, CS-) and interactions of the timeseries with these EV's with the parameters pictured below (in this example we are examining the interaction between CS+p and vmPFC timeseries- note EV's 1 and 4 are CS+p and vmPFC timeseries).

Check out the design.fsf in the example directory above for more detail on how the model was set up.