traits.trait_errors.TraitError: The 'interpolation_method' trait of a _ChooseInterpolatorOutputSpec instance must be 'LanczosWindowedSinc' or 'BSpline' or 'NearestNeighbor', but a value of 'Linear' <class 'str'> was specified.

[chiuhoward]

Summary

Interpolation fails downstream when upsampling my data by more than 10%.

Additional details

• QSIPrep version: unstable (sha256:be42cbae975ed784ac020405b97bc363213ae446568e7d2ccfbf31c88208f3a8)
• Docker version:
• Singularity version: apptainer version 1.3.2-1.el7

What were you trying to do?

My T1w data was obtained at a 1mm resolution, and my DWI images were obtained at a 2mm resolution. When I put output resolution at 1.7, the interpolation step failed. QSIprep completed successfully when I changed output resolution to 2.

What did you expect to happen?

QSIprep to complete even at output resolution of 1.7.

What actually happened?

"If you are upsampling your data by more than 10%, QSIPrep will use Linear interpolation instead of Lanczos windowed Sinc interpolation.” (https://github.com/PennLINC/qsiprep/blob/2967323a2e9634c18afa7a2b7e76048bed13a025/docs/quickstart.rst?plain=1#L97C24-L99C15) However, the ChooseInterpolatorOutputSpec step does not take Linear as a valid argument, but only 'LanczosWindowedSinc' or 'BSpline' or ‘NearestNeighbor’.

Reproducing the bug

This completed running, but the exact same code with —output-resolution 1.7 failed.

export APPTAINERENV_FS_LICENSE=$GROUP_HOME/freesurferlicense.txt
export APPTAINER_TMPDIR='/scratch/users/chiuh/tmp'
export APPTAINER_CACHEDIR='/scratch/users/chiuh/cache'

singularity run --cleanenv \
/home/groups/jyeatman/software/singularity_images/qsiprep-0.23.0-unstable-11Oct24.sif \
/scratch/groups/jyeatman/howard/DBPFullwT1Oct24/DBPFullwT1 \
/scratch/groups/jyeatman/howard/DBPFullwT1derivOct24/ participant --participant-label sub-$SLURM_ARRAY_TASK_ID \
-w /scratch/groups/jyeatman/howard/DBPFullwT1derivOct24-work/ \
--infant \
--nthreads 10 \
--omp-nthreads 8 \
--output-resolution 2 \
--unringing-method rpg \
--denoise-method dwidenoise \
--b1-biascorrect-stage none \
--pepolar-method TOPUP+DRBUDDI