Unsatisfactory result of SDC by Phase-difference B0 estimation

Hi, qsiprep experts

We are using qsiprep 0.13.0 process dMRI data with FMB SDC method(one phase difference fieldmap and two magnitude feildmaps). However, the SDC result get weird with multiple scatters outside of the brain. Any ideas what could be the reason that cause these artifacts?

Here are some results related to SDC in html.

Fieldmap to EPI registration
Results of affine coregistration between the magnitude image of the fieldmap and the reference EPI image



Fieldmap
Overlaid on the reference EPI image


Susceptibility distortion correction
Results of performing susceptibility distortion correction (SDC) on the EPI


An additional fmap/sub-001_phasediff.json file contains the info:
{
“EchoTime1”:0.00492,
“EchoTime2”:0.00738,
“IntendedFor”: “dwi/sub-test001_dwi.nii.gz”
}

Hello,

Can you highlight one of the artifacts you are talking about? This SDC correction doesn’t look too bad. If you are unsatisfied, you can try upgrading your qsiprep version and/or using the SYN fieldmapless method and see if you get better results.

Best,
Steven

Hi Steven,
Here I highlight the artifacts. The images show the artifacts which are mainly anterior for one subject and posterior for another subject. For me, this artifacts seems make the image not as clean as before.
Anterior



Posterior

Sorry for unclear illustration and thanks for your reply. I will try an updated qsiprep version.
.
Best,
Jingli

It looks like the artifact you’re concerned about is outside of the brain. There’s not much we can do about that, and it shouldn’t affect any of your downstream work that occurs inside of a brain mask.

It looks a little like your EPI reference image has some zipper artifact in it, which has made its way into the visual report. Have you tried any reconstruction workflows to see how your ODFs look?

Hi,
Thank you for your reply. I tried mrtrix_multishell_msmt reconstruction in qsiprep. Here are the results from reconstruction html.

Constrained Spherical Deconvolution
Directionally color-coded FOD peaks overlaid on the b=0 reference image.


Constrained Spherical Deconvolution
Three views of ODFs in ROIs.

Does the ODF result suggest that the artifact did not affect the downstream work? and qsiprep preprocess result was generally accpetable?

Best,
Jingli

I think these look very good! The artifact is outside the brain and even appears to be outside of the brain mask. To be absolutely sure, you could overlay the FODs on the desc-preproc_T1w image from preprocessing to make sure they align well