Missing Confounds Output in Native Space Analysis

Summary of what happened

Hello everyone,

I’m new to the community and have a couple of questions about fMRIPrep, specifically regarding native space analysis.

To potentially improve statistical power, I want to perform my analysis in native space on a dataset of 32 subjects. I’m using fMRIPrep with the --output-spaces func argument to customize the space. However, I notice that the confounds files are not being generated.

Am I missing something in the setup, or is this an expected behavior when using only the --output-spaces func flag?

Additionally, I was expecting the T1w scans would be coregistered to native space, but I only see the desc-preproc_T1w.nii.gz and corresponding .json file. Do I need to provide additional arguments for T1w-to-native space coregistration, or is this transformation not performed by fMRIPrep, meaning I should use ANTs or another tool for this purpose?

Any guidance would be appreciated!

Thank you for the help in advance.

Command Used

apptainer run \
                --cleanenv \
                --containall \
                --bind ${FMRIPREP_RESOURCES_PATH}:/resources \
                --bind ${BIDS_DIR}:/data \
                --bind ${WORK_DIR}:/work \
                --home ${OUTPUT_DIR}:/output \
                --env TEMPLATEFLOW_HOME=/resources/templateflow \
        ${FMRIPREP_RESOURCES_PATH}/fmriprep.sif /data /output \
        participant --participant_label $PARTICIPANT_LABEL \
        --nprocs 8 \
        --mem 8000 \
        --skip-bids-validation \
        --output-spaces func \
        -w /work \
        --debug all \
        --fs-license-file /resources/license.txt \
        --bids-filter-file /resources/bids_filter_file.json \

Version:

23.2.0

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

Apptainer

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

Skipping the BIDS validation.

Screenshots / relevant information:


Hi @Aybuke_Calik, and welcome to neurostars!

Please update fmriprep and try again. Also, you may perhaps want to analyze in T1w/anat space instead of func. It would still be in the space of the subject (no nonlinear warping applied, just a rigid transformation).

Best,
Steven

Dear Steven,

Thank you for your response. I tried with the updated software and still there is no confounds file. However, I couldn’t understand why confounds files are not generated for the func native space analysis.

I will consider running it in the T1w space as you suggested but I am trying to avoid additional resampling steps (from registrations etc.) as much as I can to preserve my statistical power. I have seen that some researchers refer to the T1w anatomical space as native space and others use the functional space. What is the common practice for native space analysis?

Bests,
Aybüke

Hi @Aybuke_Calik,

I am not sure what you mean by preserving statistical power in this context, so I cannot speak to that. In any case, the rigid transformation to T1w space should not introduce smoothness to the data like more complex nonlinear warps might.

Best,
Steven

Hi @Aybuke_Calik,

For more information on the --output-spaces config, please check Defining standard and nonstandard spaces where data will be resampled — fmriprep version documentation

Because we combine transformations in a single resampling step, I doubt you’ll gain any power by setting --output-spaces func, while inferences at the group level will possibly become more difficult.

However, I notice that the confounds files are not being generated.

This is highly likely because fMRIPrep did not finish successfully. Can you please upload the full log generated?

Do I need to provide additional arguments for T1w-to-native space coregistration, or is this transformation not performed by fMRIPrep, meaning I should use ANTs or another tool for this purpose?

This transformation is not performed by fMRIPrep so yes, you’ll need to use ANTs or another tool (basically FreeSurfer would be the easiest since the transformations are estimated with bbregister). I assume your processing has not been successful because you can QA this registration using the visual reports.