Weird overlay / shadow after SDC using TOPUP

[M_M]

Dear Neurostars,
we are trying to preprocess our DWI data using qsiprep, which has been great so far. We do however run into an issue with the SDC.
Any help would be appriciated!
Thank you so much!

Summary of what happened:

After qsiprep is applied to our DWI (to runs, one with 98 dirs, one with 99 dirs, both AP), we have strange “shadows” anterior and posterior (see attached gif). These seem to appear during SDC (using TOPUP, with an APPA scheme). To us, this does not look like it is intended. Do you have any guidance / ideas, what is going wrong?

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

    command = [
        "docker", "run", "--rm",
        "-v", f"{bids_dir}:/data:ro",
        "-v", f"{output_dir}:/out",
        "-v", f"{work_dir}:/work",
        "pennlinc/qsiprep:1.0.1",
        "/data", "/out", "participant",
        "--work-dir", "/work",
        "--bids-filter-file", "/data/code/BIDS_filter_file.json",
        "--nprocs", "8",
        "--mem", "16",
        "--participant-label", bids_id,
        "--unringing-method", "mrdegibbs",
        "--denoise-after-combining",
        "--output-resolution", "1.5",
        "--skip-bids-validation"
    ]

Version:

1.0.1

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

We are using docker.

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

└── ses-01
    ├── anat
    │   ├── sub-31_ses-01_T1w.json
    │   └── sub-31_ses-01_T1w.nii.gz
    ├── dwi
    │   ├── sub-31_ses-01_dir-AP_run-01_sbref.json
    │   ├── sub-31_ses-01_dir-AP_run-01_sbref.nii.gz
    │   ├── sub-31_ses-01_dir-AP_run-02_sbref.json
    │   ├── sub-31_ses-01_dir-AP_run-02_sbref.nii.gz
    │   ├── sub-31_ses-01_run-01_dwi.bval
    │   ├── sub-31_ses-01_run-01_dwi.bvec
    │   ├── sub-31_ses-01_run-01_dwi.json
    │   ├── sub-31_ses-01_run-01_dwi.nii.gz
    │   ├── sub-31_ses-01_run-02_dwi.bval
    │   ├── sub-31_ses-01_run-02_dwi.bvec
    │   ├── sub-31_ses-01_run-02_dwi.json
    │   └── sub-31_ses-01_run-02_dwi.nii.gz
    └── fmap
        ├── sub-31_ses-01_dir-AP_epi.json
        ├── sub-31_ses-01_dir-AP_epi.nii.gz
        ├── sub-31_ses-01_dir-PA_epi.json
        └── sub-31_ses-01_dir-PA_epi.nii.gz

Relevant log outputs (up to 20 lines):

Summary
Subject ID: sub-31
Structural images: 1 T1-weighted
Diffusion-weighted series: inputs 4, outputs 1
Output Name: sub-31_ses-01
Scan group: sub-31_ses-01 (PE Dir j-)
DWI Files:
/data/sub-31/ses-01/dwi/sub-31_ses-01_run-01_dwi.nii.gz
/data/sub-31/ses-01/dwi/sub-31_ses-01_run-02_dwi.nii.gz
Fieldmap type: epi
suffix: epi
epi: [‘/data/sub-31/ses-01/fmap/sub-31_ses-01_dir-AP_epi.nii.gz’, ‘/data/sub-31/ses-01/fmap/sub-31_ses-01_dir-PA_epi.nii.gz’]
Resampling targets: ACPC
Transform targets: [‘ACPC’, ‘MNI152NLin2009cAsym’]

B0 field mapping
Reports for: session 01.
A total of 2 distortion groups were included in the data data. Distortion group ‘0 -1 0 0.095910’ was represented by image 1 from sub-31_ses-01_run-02_dwi.nii.gz. Distortion group ‘0 1 0 0.095910’ was represented by image 0 from sub-31_ses-01_dir-PA_epi.nii.gz.

Reports for: session 01.
Summary
Phase-encoding (PE) direction: Anterior-Posterior
Susceptibility distortion correction: TOPUP

Screenshots / relevant information:

Not sure whether the upload of the images worked.

---

[Steven]

Hi @M_M,

Do you see the shadow on both run 1 and run 2, or is one run corrected better than the other? Is the brain mask affected by this aliasing? Does this happen on all subjects? Can you also try the TOPUP+DRBUDDI option?

Best,
Steven

[M_M]

Dear Steven.
Thank you very much for your response. I did try a few things over the weekend, so far, to no avail.

First, to answer your questions: yes, it happens for all scans (that are corrected using TOPUP), yes the mask is affected (it includes the “shadows”).
I am not quite sure what you mean by both run 1 and run 2. As I understand it, they are concatenated? Looking at the SDC plot, I see only one corrected DWI image (including run1 and run2, I assume), which includes the shadow.

To try and pinpoint the issue, I did
(1) re-run QSIprep after having removed run2 → same issue, shadow clearly visible
(2) tried DRBUDDI → no shadow, better SDC. Looking at the corrections of blip-up and blip-down, blip-down did actually look a worse after correction (distortions seemed increased + very smoothed, not sure whether this is expected).
(3) tried DRBUDDI+TOPUP: shadows appeared on the b=0 Reference Image / mask output plot.
(4) tried to change the totalreadouttime because we have partial fourier → did not help / change anything
(5) I also used TOPUP outside of QSIprep to create an unwarped version of the fieldmaps like so (using the new totalreadouttime):
topup
–imain=fmap_APPA.nii.gz
–datain=acqparams.txt
–config=b02b0.cnf
–out=topup_out
–fout=topup_field
–iout=topup_unwarped
This looks reasonable (no shadows).

I have thus not found a solution to the issue. As it is now, we could probably use DRBUDDI, but I would prefer TOPUP if possible (also because I would really like to understand the issue).

Thank you very much for your help!
Best,
M_M