XCP-D misaligned MNI atlas and func image + running in T1w space?

chrisvriend · September 16, 2025, 8:36am

Summary of what happened:

Dear experts,

I’m trying to use XCP-D (0.11.1) to denoise my functional images that have been preprocessed using fmrirprep and subsequently extract timeseries using one of the XCP atlases. However, the alignment of the atlases in the xcp_outputfolder/atlases (e.g. atlas-Glasser) - which I assume are being used to extract the timeseries with - is quite poor. I’m confident that this is not an issue with fmriprep because my functional image and the MNI152NLin6Asym_res-2 brain from templateflow overlap perfectly. In fact, the atlases itself do not overlap well with the MNI152NLin6Asym_res-2 template. I see this for both the Glasser and Gordon atlas (see screenshot - I’m only allowed to upload one by the system). What could be the cause of this?

Is it also possible to perform the denoising in T1w space? I tried using a bids filter to select the T1w space images in the func folder of fmriprep but that threw an error.

thanks, Chris

Command used (and if a helper script was used, a link to the helper script or the command generated):

apptainer run \
    $XCPD_IMAGE \
    ${fmriprepdir} ${outputdir} participant \
    --input-type fmriprep \
    --dummy-scans 3 \
    --nuisance-regressors 36P \
    --despike \
    --file-format nifti \
    --participant-label ${subj} \
    --work-dir ${workdir} \
    --mode linc \
    --atlases Glasser

Version:

0.11.1

Environment (Docker, Singularity / Apptainer, custom installation):

apptainer

Data formatted according to a validatable standard? Please provide the output of the validator:

PASTE VALIDATOR OUTPUT HERE

Relevant log outputs (up to 20 lines):

PASTE LOG OUTPUT HERE

Screenshots / relevant information:

fmriprep output tree

sub-1050
├── anat
│   ├── sub-1050_desc-aparcaseg_dseg.nii.gz
│   ├── sub-1050_desc-aseg_dseg.nii.gz
│   ├── sub-1050_desc-brain_mask.json
│   ├── sub-1050_desc-brain_mask.nii.gz
│   ├── sub-1050_desc-preproc_T1w.json
│   ├── sub-1050_desc-preproc_T1w.nii.gz
│   ├── sub-1050_dseg.nii.gz
│   ├── sub-1050_from-fsnative_to-T1w_mode-image_xfm.txt
│   ├── sub-1050_from-MNI152NLin2009cAsym_to-T1w_mode-image_xfm.h5
│   ├── sub-1050_from-MNI152NLin6Asym_to-T1w_mode-image_xfm.h5
│   ├── sub-1050_from-T1w_to-fsnative_mode-image_xfm.txt
│   ├── sub-1050_from-T1w_to-MNI152NLin2009cAsym_mode-image_xfm.h5
│   ├── sub-1050_from-T1w_to-MNI152NLin6Asym_mode-image_xfm.h5
│   ├── sub-1050_hemi-L_inflated.surf.gii
│   ├── sub-1050_hemi-L_midthickness.surf.gii
│   ├── sub-1050_hemi-L_pial.surf.gii
│   ├── sub-1050_hemi-L_smoothwm.surf.gii
│   ├── sub-1050_hemi-R_inflated.surf.gii
│   ├── sub-1050_hemi-R_midthickness.surf.gii
│   ├── sub-1050_hemi-R_pial.surf.gii
│   ├── sub-1050_hemi-R_smoothwm.surf.gii
│   ├── sub-1050_label-CSF_probseg.nii.gz
│   ├── sub-1050_label-GM_probseg.nii.gz
│   ├── sub-1050_label-WM_probseg.nii.gz
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_desc-brain_mask.json
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_desc-brain_mask.nii.gz
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_desc-preproc_T1w.json
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_desc-preproc_T1w.nii.gz
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_label-CSF_probseg.nii.gz
│   ├── sub-1050_space-MNI152NLin6Asym_res-2_label-GM_probseg.nii.gz
│   └── sub-1050_space-MNI152NLin6Asym_res-2_label-WM_probseg.nii.gz
├── func
│   ├── aroma_motion_regressors.txt
│   ├── sub-1050_task-rest_AROMAnoiseICs.csv
│   ├── sub-1050_task-rest_desc-brain_mask.json
│   ├── sub-1050_task-rest_desc-brain_mask.nii.gz
│   ├── sub-1050_task-rest_desc-confounds_timeseries.json
│   ├── sub-1050_task-rest_desc-confounds_timeseries.tsv
│   ├── sub-1050_task-rest_desc-MELODIC_mixing.tsv
│   ├── sub-1050_task-rest_desc-preproc_bold.json
│   ├── sub-1050_task-rest_from-scanner_to-T1w_mode-image_xfm.txt
│   ├── sub-1050_task-rest_from-T1w_to-scanner_mode-image_xfm.txt
│   ├── sub-1050_task-rest_space-fsaverage5_hemi-L_bold.func.gii
│   ├── sub-1050_task-rest_space-fsaverage5_hemi-L_bold.json
│   ├── sub-1050_task-rest_space-fsaverage5_hemi-R_bold.func.gii
│   ├── sub-1050_task-rest_space-fsaverage5_hemi-R_bold.json
│   ├── sub-1050_task-rest_space-fsnative_hemi-L_bold.func.gii
│   ├── sub-1050_task-rest_space-fsnative_hemi-L_bold.json
│   ├── sub-1050_task-rest_space-fsnative_hemi-R_bold.func.gii
│   ├── sub-1050_task-rest_space-fsnative_hemi-R_bold.json
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.json
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.nii.gz
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_boldref.nii.gz
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-aparcaseg_dseg.nii.gz
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-aseg_dseg.nii.gz
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-brain_mask.json
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-brain_mask.nii.gz
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-preproc_bold.json
│   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-preproc_bold.nii.gz
│   ├── sub-1050_task-rest_space-T1w_boldref.nii.gz
│   ├── sub-1050_task-rest_space-T1w_desc-aparcaseg_dseg.nii.gz
│   ├── sub-1050_task-rest_space-T1w_desc-aseg_dseg.nii.gz
│   ├── sub-1050_task-rest_space-T1w_desc-brain_mask.json
│   ├── sub-1050_task-rest_space-T1w_desc-brain_mask.nii.gz
│   ├── sub-1050_task-rest_space-T1w_desc-preproc_bold.json
│   └── sub-1050_task-rest_space-T1w_desc-preproc_bold.nii.gz

xcp-output tree

atlases
│   ├── atlas-4S1056Parcels
│   │   ├── atlas-4S1056Parcels_dseg.tsv
│   │   ├── atlas-4S1056Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S1056Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S156Parcels
│   │   ├── atlas-4S156Parcels_dseg.tsv
│   │   ├── atlas-4S156Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S156Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S256Parcels
│   │   ├── atlas-4S256Parcels_dseg.tsv
│   │   ├── atlas-4S256Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S256Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S356Parcels
│   │   ├── atlas-4S356Parcels_dseg.tsv
│   │   ├── atlas-4S356Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S356Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S456Parcels
│   │   ├── atlas-4S456Parcels_dseg.tsv
│   │   ├── atlas-4S456Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S456Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S556Parcels
│   │   ├── atlas-4S556Parcels_dseg.tsv
│   │   ├── atlas-4S556Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S556Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S656Parcels
│   │   ├── atlas-4S656Parcels_dseg.tsv
│   │   ├── atlas-4S656Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S656Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S756Parcels
│   │   ├── atlas-4S756Parcels_dseg.tsv
│   │   ├── atlas-4S756Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S756Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S856Parcels
│   │   ├── atlas-4S856Parcels_dseg.tsv
│   │   ├── atlas-4S856Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S856Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-4S956Parcels
│   │   ├── atlas-4S956Parcels_dseg.tsv
│   │   ├── atlas-4S956Parcels_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-4S956Parcels_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-Glasser
│   │   ├── atlas-Glasser_dseg.tsv
│   │   ├── atlas-Glasser_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-Glasser_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-Gordon
│   │   ├── atlas-Gordon_dseg.tsv
│   │   ├── atlas-Gordon_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-Gordon_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-HCP
│   │   ├── atlas-HCP_dseg.tsv
│   │   ├── atlas-HCP_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-HCP_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   ├── atlas-Tian
│   │   ├── atlas-Tian_dseg.tsv
│   │   ├── atlas-Tian_space-MNI152NLin6Asym_res-2_dseg.json
│   │   └── atlas-Tian_space-MNI152NLin6Asym_res-2_dseg.nii.gz
│   └── dataset_description.json
├── dataset_description.json
├── desc-linc_qc.json
├── logs
│   ├── CITATION.bib
│   ├── CITATION.html
│   ├── CITATION.md
│   └── CITATION.tex
├── sub-1050
│   ├── anat
│   │   └── sub-1050_space-MNI152NLin6Asym_desc-preproc_T1w.nii.gz
│   ├── func
│   │   ├── sub-1050_task-rest_desc-preproc_design.json
│   │   ├── sub-1050_task-rest_desc-preproc_design.tsv
│   │   ├── sub-1050_task-rest_motion.json
│   │   ├── sub-1050_task-rest_motion.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_desc-linc_qc.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-denoised_bold.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-denoised_bold.nii.gz
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-denoisedSmoothed_bold.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-denoisedSmoothed_bold.nii.gz
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-alff_boldmap.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-alff_boldmap.nii.gz
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-alff_desc-smooth_boldmap.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-alff_desc-smooth_boldmap.nii.gz
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-reho_boldmap.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_res-2_stat-reho_boldmap.nii.gz
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-alff_bold.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-alff_bold.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-coverage_bold.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-coverage_bold.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-mean_timeseries.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-mean_timeseries.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-pearsoncorrelation_relmat.json
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-pearsoncorrelation_relmat.tsv
│   │   ├── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-reho_bold.json
│   │   └── sub-1050_task-rest_space-MNI152NLin6Asym_seg-Glasser_stat-reho_bold.tsv
├── sub-1050.html

tsalo · September 16, 2025, 1:33pm

Can you please add the XCP-D command you used to your initial post?

Can you share the tree output for both your fMRIPrep and XCP-D outputs?

No, native-space processing isn’t supported in XCP-D yet.

chrisvriend · September 16, 2025, 1:56pm

I’ve added the tree and code to the original post. thanks!

tsalo · September 16, 2025, 4:25pm

Thank you. What files did you use for the image you created? Can you paste in the full paths to those files?

chrisvriend · September 16, 2025, 6:46pm

you mean of the screenshot I sent? those were
xcp_out/atlases/atlas-Gordon/atlas-Gordon_space-MNI152NLin6Asym_res-2_dseg.nii.gz overlaid on
../templateflow/tpl-MNI152NLin6Asym/tpl-MNI152NLin6Asym_res-02_desc-brain_T1w.nii.gz
but when using either
xcp_out/sub-1050/func/sub-1050_task-rest_space-MNI152NLin6Asym_res-2_desc-denoised_bold.nii.gz
or
xcp_out/sub-1050/anat/sub-1050_space-MNI152NLin6Asym_desc-preproc_T1w.nii.gz
the misalignment is the same. Also when using the Glasser atlas (but then more visible in the parietal cortex)

tsalo · September 16, 2025, 8:33pm

I reproduced the MNI152NLin6Asym res-2 atlas locally and it looks the same. I think it’s just not that well-registered to begin with, since the atlas is already in MNI152NLin6Asym and XCP-D is just downsampling it to 2x2x2 mm voxels.

I also checked the 1 mm3 versions of the atlas and template, and the same level of mismatch is apparent.

chrisvriend · September 17, 2025, 10:22am

Ok, so does that mean that we should not use the timeseries or matrices that are extracted by XCP-D for these atlases?

tsalo · September 17, 2025, 12:40pm

That’s entirely up to you. The registration seems reasonable in most of the brain, but there are definitely areas where it’s not great. FWIW I think most people use the CIFTI version of this atlas.