Qsiprep Preprocessing Issue

1

There seems to be an issue with qsiprep’s output. The eddy and topup-corrected output doesn’t look preprocessed (see figures below).

Relevant info:

  • qsiprep version: 0.18.0alpha0
  • qsiprep command: /usr/local/miniconda/bin/qsiprep /path/to/BIDS/ /path/to/BIDS/derivatives/ participant --participant-label sub-X -w /path/to/qsiprep/work --fs-license-file /path/to/license.txt --hmc-transform Rigid --pepolar-method TOPUP --output-resolution 2

image
image1666×1700 58.3 KB

image
image1666×1220 54.9 KB

Is qsiprep being run correctly with the above command?

2

Hi @ameenq05,

You should use the latest version to ensure there’s not an issue that’s already been addressed. What in particular are you worried about? If it’s the distortions, have you confirmed SDC was performed? Is it possible phase encoding directions were not correctly specified?

Best,
Steven

3

Hi Steven,

Thank you for your response. I’ve processed the same dataset using the latest version of qsiprep (0.21.5.dev0+g36b93fe.d20240504), but the distortions and noisiness persist.

I am concerned with the frontal lobe distortions, and the noisiness of the data overall. I can confirm that the phase encoding directions were specified and recognized by qsiprep, with SDC being performed (confirmed by looking at the .html report). Despite SDC and dwidenoise being performed on the data, the output diffusion image is still noisy and SDC seems incomplete.

Best,
Ameen