Bad preprocessing results for dwi with PSR+ image orientation

[jhauneuro]

Summary

I am trying to preprocess a 2D EPI dwi image acquired sagittally. My preprocessed data looks bad (i.e., it has dark and bright spots that don't correspond to the anatomy) when run on my dwi image oriented PSR+ (converted using dcm2niix). When I reorient the image to LAS+ and rerun the preprocessing the preprocessed image looks much better.

However, when I reorient the image to LAS+ I am not able to use eddy's intra-volume motion correction since the slice timing info in the json file no longer corresponds to the reoriented nifti image. This is too bad because there is quite a bit of motion in the data.

Does qsiprep require the data to have a specific image orientation? What could explain the different results?

Additional details

• QSIPrep version: 0.19.1 (unable to run this in 0.22.0 due to issue #792
• Docker version: 25.0.4

What were you trying to do?

Preprocess my dwi data.

What did you expect to happen?

Successfully complete preprocessing.

What actually happened?

Preprocessed results look bad (bright and dark spots appear in areas where they shouldn't).

Reproducing the bug

Here is the command I ran:

docker run -ti --rm \
	-v /usr/local/freesurfer/7.4.1/license.txt:/opt/freesurfer/license.txt:ro \
	-v $bids_dir:/data:ro \
	-v $eddy_config_file:/sngl/eddy/eddy_config.json:ro \
	-v $out_dir:/out \
	-v $out_dir:/scratch \
	pennbbl/qsiprep:0.19.1 /data /out participant \
	--eddy-config /sngl/eddy/eddy_config.json \
	--output-resolution $res --pepolar-method TOPUP \
	--bids-database-dir $bids_dir --anat-modality none \
	-w /scratch --omp-nthreads 1 --nthreads 1 --participant_label $subjID

Screenshots of the final preprocessed b0 reference image when the input dwi is oriented in:

PSR+

LAS+

[mattcieslak]

I've never run into PSR dwi data before, so it's possible there is something missing in the workflow that reads in the images. Would you be able to share some of this PSR data?

[jhauneuro]

@mattcieslak do you have access to the NIMH Data Archive? This PSR data comes from NDA Collection 3088. You'd need to have a non-expired Data Use Certificate for me to share it with you.