Issues with Combining Multi-shell DWI Data for NODDI Analysis in OpenNeuro ds003138 Dataset

1

Hello everyone,

I would like to perform a study using the diffusion data from the OpenNeuro dataset “Tidying Up White Matter: Neuroplastic Transformations in Sensorimotor Tracts following Slackline Skill Acquisition” (OpenNeuro)).

In this dataset, the diffusion-weighted images (DWI) are organized in the dwi folder with shell_1 = b1000, shell_2 = b2000, and shell_3 = b3000. My aim is to combine these shells and generate NODDI maps.

I merged the shells using fslmerge and then performed preprocessing with QSIPrep. However, the resulting NODDI and diffusion kurtosis images did not look correct. I also tried preprocessing with FSL eddy (without using QSIPrep), but encountered similar issues.

I could not find a clear explanation for the source of this problem or a recommended solution.

I would greatly appreciate any insights or advice you might have.
Thank you very much for your help!

2

Hi @mdbaris and welcome to neurostars!

Qsiprep merges DWI automatically. I wouldn’t touch the raw data before working with Qsiprep as long as its bids valid.

It would also help to see your code and screenshots of what looks wrong.

Best,
Steven

3

Hi Steven,

Thank you very much for your quick response!

I did not modify or manually merge the raw DWI data.
I directly ran QSIPrep on the original dataset without any changes.

However, QSIPrep automatically merged shell-2 (b=2000) and shell-3 (b=3000) together, but shell-1 (b=1000) was not included in the combined output.
I carefully checked the filenames and BIDS structure, and they seem fully BIDS-compliant.

For reference, here is the QSIPrep command I used (running via Docker):

docker run -ti --rm --gpus all \
  -v /mnt/c/Users/baris/documents/egzersiz/ds003138-download:/data:ro \
  -v /mnt/c/Users/baris/documents/egzersiz/qsiprep_out:/out \
  -v /mnt/c/Users/baris/documents/egzersiz/license.txt:/opt/freesurfer/license.txt \
  pennlinc/qsiprep:latest /data /out participant \
  --fs-license-file /opt/freesurfer/license.txt \
  --output-resolution 1.2 \
  --eddy-config /mnt/c/Users/baris/documents/egzersiz/eddy_params.json

I am also attaching screenshots showing the outputs and the shell merging issue.

Thank you again for your support!

Best regards,
Barış

Ekran görüntüsü 2025-04-27 152039
Ekran görüntüsü 2025-04-27 152039784×496 28.5 KB

4

Hmm that may be a bug, @mattcieslak what do you think?

Can you try for one subject just having the DWIs as run-1 run-2 and run-3 with no acq label? Keep in mind you’ll have to change the IntendedFor for the fmaps to be linked properly.

Best,
Steven