3dMVM error: datasets are not on the same grid

AFNI version: afni 25.1.10

Hi all!

I am interested in how changing hormones ("FSH" - quantitative, between-subjects) in the menopause transition relate to seed-connectivity and wrote a script using 3dMVM to investigate this:

3dMVM -prefix MVM_FSH_peri.nii.gz -jobs 4   \
      -mask /data1/software/bioimagesuite35/images/MNI_T1_1mm_mask.nii.gz \
      -bsVars  'FSH'      \
      -qVars    'FSH'     \
      -qVarCenters   '13.2'       \
      -SS_type 3          \
      -num_glt 1 \
      -gltLabel 1 'FSH-Zscore'   -gltCode 1 'FSH : '    \
      -resid MVM_FSH_peri_resid.nii.gz   \
      -dataTable      \
Subj    FSH     InputFile   \

It gives me an error that the datasets are not on the same grid:

Dimensions: 
rows: 48 | columns: 3

Data summary: 
Variable  Detected_Type Details                               
Subj      Subjects      Num Subjects=48                       
FSH       Quantitative  Min=0.81 | Max=61.56 | Num outliers=5 
InputFile Data          Number of InputFiles=48               

++ Good: All InputFiles exist.

++ Good: All InputFiles have exactly 1 volume.

** ERROR: Datasets are NOT on the same grid!!!
-- Note: Grids are compared to the 1st InputFile.

When I do 3dinfo -ad3 -n4 -dc3, all of my input files populate like this:

** WARNING: NIfTI file HCA6198582_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
1.000000	1.000000	1.000000	181	217	181	1	-13.344810	-9.775955	42.264732	HCA6198582_V1_MR_bis_seed_1_MNI151.nii.gz
** WARNING: NIfTI file HCA6439780_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
1.000000	1.000000	1.000000	181	217	181	1	-13.291008	-25.699692	88.080994	HCA6439780_V1_MR_bis_seed_1_MNI151.nii.gz
** WARNING: NIfTI file HCA6742377_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
1.000000	1.000000	1.000000	181	217	181	1	-19.238884	-15.837227	64.392929	HCA6742377_V1_MR_bis_seed_1_MNI151.nii.gz

So it seems like the problem is that the geometric centers are off but I am not sure how, since the matrix and voxel sizes are the same. All the data were collected from the HCP-A project and preprocessed in the same pipeline. I also don't know if that Warning that prints out matters, since that has printed in many previous analyses from my lab and I was told to ignore it. Please let me know how else to investigate this error and what I can do to fix it!

Jordan

Also, for bonus points - I originally ran this analysis using 3dLME and specifying -ranEff '~1' as had been done in my lab with similar analyses, which ran completely fine. However, I was told here to use 3dMVM since there are no within-subjects effects and I am wondering if someone could comment on my lab's justification for doing this from a prior study and if running it as 3dLME vs 3dMVM would be expected to have different outcomes: We have clarified in the revision that we did use a regression and not linear mixed effects modeling. While AFNI’s 3dLME is an excellent tool for linear mixed effects modeling, in a case (such as this one) without repeated measurements, we also find it to be a straightforward tool for modeling linear regressions.

Hi Jordan,

Would you please try "3dinfo -obliquity -space" on those datasets? If they are differently oblique, they would not be on the same grid. What standard space are these datasets on?

-rick

Hi Rick!

These are all in MNI 151 space. Here are the results from "3dinfo -obliquity -space":

** WARNING: NIfTI file HCA6198582_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
11.307	TLRC
** WARNING: NIfTI file HCA6439780_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
22.684	TLRC
** WARNING: NIfTI file HCA6742377_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added
13.566	TLRC

So I guess they are differently oblique. These input files are seed-connectivity images from an analysis ran through bioimage suite but I get the same findings when doing the same with the preprocessed images before seed-analysis. Is this problem from a preprocessing step (or lack thereof) and how would I adjust this now?

Jordan

That warning is suspicious:

** WARNING: NIfTI file HCA6198582_V1_MR_bis_seed_1_MNI151.nii.gz dimensions altered since AFNI extension was added

You might try removing the old and possibly inappropriate extensions by running:

nifti_tool -rm_ext ALL -infile XXXX.nii.gz -prefix XXXX_noext.nii.gz

on each file. Then test with the _noext versions.
-rick

Removing the extensions does get rid of that warning, but the obliquities are unchanged and script is still outputting the error that the datasets are not on the same grid. I also tried to remove from the preprocessed images pre-analyses and the obliquities are still different from one another.

Jordan

Hey Rick and anyone seeing this! I've also been asking around my lab group about this obliquity error and it seems like the problem may be that I used bioimage suite to reslice my images to MNI space, so 3dMVM works with images resliced with a different command and I'll just have to rerun those steps for this 3dMVM. Thank you for your help!!

Jordan

That's great, Jordan, thanks for the update!

-rick