Hi Donald and others,

I am trying to run a gPPI with one task regressor (off vs. on) which is modulated by three parametric regressors. AKA on each trial ("task on") the HRF is modulated by three pm modulators. Others here in the forum tried it too and made it work. So it should be possible, technically. I tried to use the advice that was mentioned in those threads. gPPI is running: setting up the design matrix, estimating the hyperparamters writing the SPM.mat. But then an SPM error message pops up: "No inmask voxels - empty analysis!".

The mask image on ss level for that particular person includes the seed voxel. Furthermore the seed voxel stems from a second level analysis so the seed voxel must be present in all subjects by definition.

In P.maskdir I have the group mask image of a second level analysis. It also includes the VOI. Hence: P.VOI.masks -> ans=cell('mask.img'). I tried also with P.FLmask =1 and P.FLmask =0; No difference; Leaving P.VOI.masks empty does not help; it leads to gPPI not doing anything anymore.

This is my P structure:

subject: 'BGG_3009'
VOI: [1x1 struct]
Region: 'DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask_cond'
SPMver: 8
directory: 'F:\AG_Diplomarbeit\Preprocessed_on_Windows_Machine_swuaf_VBM8\AD\BGG_3009\results_abs_loss'
Estimate: 1
contrast: {'Omnibus F-test for PPI Analyses'}
extract: 'eig'
Tasks: {'0' 'Task_on'}
Weights: []
maskdir: 'F:\AG_Diplomarbeit\Preprocessed_on_Windows_Machine_swuaf_VBM8\batch_templates\SS_gPPI\maskdir'
equalroi: 0
FLmask: 1
VOI2: {}
analysis: 'psy'
method: 'cond'
CompContrasts: 1
Weighted: []
Contrasts: [1x2 struct]
VOImin: 1

This is the output:

SUBJECT LIST:
Adding BGG_3009 to subject list
Adding BGG_3031 to subject list
Adding BGG_3032 to subject list
Adding BGG_3036 to subject list
Adding BGG_3038 to subject list
Adding BGG_3039 to subject list
Adding BGG_3041 to subject list
Adding BGG_3047 to subject list
Adding BGG_3055 to subject list
Adding BGG_3057 to subject list
Adding BGG_3059 to subject list
Adding BGG_3060 to subject list

MASK LIST:
Adding DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii to mask list
Log File: BGG_3009_PPPI_12_8_2014_23.log
PPPI Version: 13.1.4-17-2014
Parameters used:
Processing subject: BGG_3009
VOI file : F:\AG_Diplomarbeit\Preprocessed_on_Windows_Machine_swuaf_VBM8\batch_templates\SS_gPPI\VOIs\DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii
Output file will be: BGG_3009_DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask_cond_session#_cond_PPI_regressors.txt
Contrast : Omnibus F-test for PPI Analyses
Analysis : Psychophysiological Interactions
Extraction : eigenvariate
Tasks : _0_Task_on
Method : Condition Specific
------------------------------------------------------------------------
Running job #1
------------------------------------------------------------------------
Running 'Volume of Interest'
## Note: this VOI facility is in a beta version. ##
## Interface and features might change in the future. ##
VOI saved as ..-40_46_24_HC_gr_PAT_neural_LA_mask_BGG_3009_Sess_1.mat
Done 'Volume of Interest'
Done
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space. This is in the the same space as the input data and functional mask.
SPM8: spm_fMRI_design (v4185) 15:23:19 - 08/12/2014
========================================================================
Saving fMRI design : ...SPM.mat saved
SPM8: spm_fmri_spm_ui (v4421) 15:23:19 - 08/12/2014
========================================================================
Mapping files : ...done
Calculating globals : ...done
Saving SPM configuration : ...SPM.mat saved
Design reporting : ...done
estimate_PPI.m
SPM8: spm_spm (v4515) 15:23:21 - 08/12/2014
========================================================================
Initialising parameters : ...done
Plane 39/39 , block 1/1 : ...done
Valid Contrast
Valid Contrast
PPI Contrasts were not estimated for some reason.

I hope someone can help!

Thanks,
Alex

That's an odd error message. Could you please send me the SPM.mat file, the
timeseries images, and the SS mask file, and the group mask file. As the
data worked on the standard task activity analysis, it should also work
here. Does this happen on other subjects or just this one subject?

Hi Donald,

here's the SPM.mat. (from the gPPI analysis).

Here's the mask.img from ss level

Here's the mask.img from one of the second level analyses. This one I also tried to use for masking the seed VOI.

This is the SPM.mat from the original ss model in the subject.

It happens on all subjects. I have noticed now, that when I ran it on a subject that I have never had run it on before I get this output:

MASK LIST:
Adding DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii to mask list
Log File: BGG_3032_PPPI_12_8_2014.log

PPPI Version: 13.1.4-17-2014
Parameters used:
Processing subject: BGG_3032
VOI file : F:\AG_Diplomarbeit\Preprocessed_on_Windows_Machine_swuaf_VBM8\batch_templates\SS_gPPI\VOIs\DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii
Output file will be: BGG_3032_DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask_cond_session#_cond_PPI_regressors.txt
Valid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Valid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Valid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Valid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
ESS image 10 : ...written ess_0010.img
spm{F} image 10 : ...written spmF_0010.img
Contrast : Omnibus F-test for PPI Analyses
Analysis : Psychophysiological Interactions
Extraction : eigenvariate
Tasks : _0_Task_on
Method : Condition Specific
------------------------------------------------------------------------
Running job #1
------------------------------------------------------------------------
Running 'Volume of Interest'
## Note: this VOI facility is in a beta version. ##
## Interface and features might change in the future. ##
VOI saved as ..-40_46_24_HC_gr_PAT_neural_LA_mask_BGG_3032_Sess_1.mat
Done 'Volume of Interest'
Done
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space. This is in the the same space as the input data and functional mask.
SPM8: spm_fMRI_design (v4185) 18:31:23 - 08/12/2014
========================================================================
Saving fMRI design : ...SPM.mat saved
SPM8: spm_fmri_spm_ui (v4421) 18:31:23 - 08/12/2014
========================================================================
Mapping files : ...done
Calculating globals : ...done
Saving SPM configuration : ...SPM.mat saved
Design reporting : ...done
estimate_PPI.m
SPM8: spm_spm (v4515) 18:31:26 - 08/12/2014
========================================================================
Initialising parameters : ...done
Plane 39/39 , block 1/1 : ...done
Valid Contrast
Valid Contrast
PPI Contrasts were not estimated for some reason.

Yet if I run it again on that subject I get this:

SUBJECT LIST:
Adding BGG_3009 to subject list
Adding BGG_3031 to subject list
Adding BGG_3032 to subject list
Adding BGG_3036 to subject list
Adding BGG_3038 to subject list
Adding BGG_3039 to subject list
Adding BGG_3041 to subject list
Adding BGG_3047 to subject list
Adding BGG_3055 to subject list
Adding BGG_3057 to subject list
Adding BGG_3059 to subject list
Adding BGG_3060 to subject list

MASK LIST:
Adding DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii to mask list
Log File: BGG_3032_PPPI_12_8_2014_3.log
PPPI Version: 13.1.4-17-2014
Parameters used:
Processing subject: BGG_3032
VOI file : F:\AG_Diplomarbeit\Preprocessed_on_Windows_Machine_swuaf_VBM8\batch_templates\SS_gPPI\VOIs\DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask.nii
Output file will be: BGG_3032_DLPFC_-40_46_24_HC_gr_PAT_neural_LA_mask_cond_session#_cond_PPI_regressors.txt
Contrast : Omnibus F-test for PPI Analyses
Analysis : Psychophysiological Interactions
Extraction : eigenvariate
Tasks : _0_Task_on
Method : Condition Specific
------------------------------------------------------------------------
Running job #1
------------------------------------------------------------------------
Running 'Volume of Interest'
## Note: this VOI facility is in a beta version. ##
## Interface and features might change in the future. ##
VOI saved as ..-40_46_24_HC_gr_PAT_neural_LA_mask_BGG_3032_Sess_1.mat
Done 'Volume of Interest'
Done
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space
VOI has 1 voxels in 3.500000e+000x3.500000e+000x3.500000e+000 space. This is in the the same space as the input data and functional mask.
SPM8: spm_fMRI_design (v4185) 18:45:00 - 08/12/2014
========================================================================
Saving fMRI design : ...SPM.mat saved
SPM8: spm_fmri_spm_ui (v4421) 18:45:00 - 08/12/2014
========================================================================
Mapping files : ...done
Calculating globals : ...done
Saving SPM configuration : ...SPM.mat saved
Design reporting : ...done
estimate_PPI.m
SPM8: spm_spm (v4515) 18:45:02 - 08/12/2014
========================================================================
Initialising parameters : ...done
Plane 39/39 , block 1/1 : ...done
Valid Contrast
Valid Contrast
PPI Contrasts were not estimated for some reason.

This is the wrapper script I am using (originally by Bob Spunt)

I have noticed also, that when I am using the wrapper script I am losing spm defaults:

>> spm('CheckModality')
Modality is not set: use spm('defaults','MOD'); where MOD is one of PET, FMRI, EEG.
??? Error using ==> spm at 495
Unknown Modality.

This becomes noticable when I want to display results using an spm.mat on 2nd level, for example, I cannot do it anymore. I first have to set the modality again.

spm('defaults','fMRI')

Yet when I just use PPPI(P) as a command (without wrapper), then I still get the error message 'no inmask voxels' but I don't lose the spm defaults. I don't know if this is just a symptom of the problem or part of the reason.

The time series data (preprocessed niftis of the one subject, 42 MB) I am uploading right now to a public downloadable link. When it is done, I'll post the link here!

Thank you for looking into it!

Also when chaning in the wrapper script, starting at line 175:

if strcmp(method,'cond')

% USER INPUT: Specify your contrasts of interest (Cond PPI only)
%--------------------------------------------------------------%
P.Contrasts(1).name = 'gain';
P.Contrasts(1).left = {'Task_on'}; %left is always bigger than right
P.Contrasts(1).Contrail.L = {'xgain^1'};
P.Contrasts(1).right = {''};
P.Contrasts(1).STAT = 'T';
P.Contrasts(1).MinEvents = 5;
%--------------------------------------------------------------%
end

... the error still remains.

Also when using instead of contrail "xgain^1" just "gain", as can be found in SPM.Sess([1]).U(1).P.name (in the SPM.mat) written in the PPPI process.

Here's the link to the time series data

https://drive.google.com/file/d/0B4OSCzz...

Best,
Alex

For those of you who are following, the error was that the filenames for the scans were incorrect in the gPPI model.

The 
"Valid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast"

message that repeats is expected. This means that you do not have temporal or dispersion derivatives in the model. If you have derivatives, then you will not see the invalid contrast warning messages.

Thanks a lot for the help! It works now.
 
Where could I read about the possibilities of Dispersion derivatives in the model?
 
Best,
Alex