Distortion correction via phase difference image produces oddly distorted output epis


We just started using fmriprep (on Siemens 7T data), which does a great job. Only the B0 correction produces oddly distorted results for all subjects.

We aim to perform sdc the “old way” using a phase difference image plus one (of the two) magnitude image(s), which we provide (without further preprocessing) in bids format: fmap/sub-001_phasediff.nii.gz and fmap/sub-001_magnitude1.nii.gz, respectively.

An additional fmap/sub-001_phasediff.json file contains the info:

“EchoTime1”: 0.00600,
“EchoTime2”: 0.00702,
“IntendedFor”: “func/sub-001_task-heat_bold.nii.gz”

Is it correct that fmriprep wants the phase difference /magnitude image without further preprocessing (unlike e.g., FUGUE)? Any hints are very much appreciated.

Thank you for your help,

phase difference /magnitude image input data:

This issue was solved by copying the json files generated by dcm2niix to the phase difference image (+ “IntendedFor”), the magnitude image and the epi time series. I didn’t check on exactly which info in the json files is necessary to perform a successful sdc. Feel free to add to it if you know. I leave this post for illustration.

Hi, I’m having the same issue with SDC. can you clarify what json files you copied and to where?

Chris Rordan’s dcm2niix tool creates a corresponding json file for every nii. SDC worked for me when providing the original json files for the fieldmaps (magn & phase) & epi timeseries – file naming & folder structure according to BIDS convention. It didn’t work before in my hands bc I edited the content of the original json files.