search

nipreps / fmriprep

fMRIPrep is a robust and easy-to-use pipeline for preprocessing of diverse fMRI data. The transparent workflow dispenses of manual intervention, thereby ensuring the reproducibility of the results.
https://fmriprep.org
Apache License 2.0
613 stars 288 forks source link

Susceptibility distortion correction (SDC) using Spin Echo Fieldmaps #2210

Open sameera2004 opened 4 years ago

sameera2004 commented 4 years ago

Hi,

I pre-processed resting state data using fMRIprep 20.1.1. This data set contain 20 subjects, 2 sessions for each subject, and 4 runs for each session. I used SpinEcho field maps to do the susceptibility distortion correction (SDC). The SDC with SpinEcho’s worked for majority of the data. However, it worsen the SDC’s in certain subjects. Below is the command that I used to run fMRIprep;

singularity run --cleanenv --bind /mnt/fMRIprep /mnt/tools/fMRIprep_sing/fmriprep-20.1.1.simg /mnt/fMRIprep/scR21 /mnt/fMRIprep/scR21_outputs -w /mnt/fMRIprep/scR21_outputs/derivatives/scratch --fs-license-file $PWD/license.txt --write-graph --cifti-output 91k participant --participant-label 50015 50016 50020 50024 50027 50033 50042 50137 50144 --ignore slicetiming --output-spaces T1w MNI152NLin2009cAsym:res-2 fsaverage fsLR I am wondering if anybody else has experienced similar issues. Please let me know.

Thanks in advance Best Regards Sameera

Screen Shot 2020-07-07 at 5 20 14 PM Screen Shot 2020-07-07 at 5 20 25 PM
oesteban commented 2 years ago

The reportlet and the preproc nifti indeed have divergent paths in the compute graph.

Bad news is that a fix requires a much more comprehensive implementation of resampling on SDCFlows (e.g., 4D files, accept head-motion parameters). Good news is that this new implementation will stick out w.r.t. the state of the art, leveling with FSL [much lesser flexible] implementations and, to the best of my knowledge, much better than any other tool.

This is to say this will take a little time, but both dMRIPrep and fMRIPrep are going to benefit great deal.

akhanf commented 2 years ago

Just bumping this as having the same problems when processing data with HCP Lifespan protocol -- wondering if there is a potential fix for this in some development branch? Thanks!

oesteban commented 2 years ago

@akhanf thanks for the nudge, indeed there's a fix recently merged in master. That means that the nipreps/fmriprep:unstable docker image should be up-to-date with the fix.

@sameera2004 any chance you could give the new Docker image a try with these data?

akhanf commented 2 years ago

Great to hear, thanks @oesteban ! Will give it a try too..