-
Notifications
You must be signed in to change notification settings - Fork 11
Reinvent dmriprep? #103
base: master
Are you sure you want to change the base?
Reinvent dmriprep? #103
Conversation
Hello @dPys! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:
Comment last updated at 2019-09-08 07:57:18 UTC |
And if anyone wishes to experiment, a quick and dirty run can be accomplished as follows:
|
correct hemi check correct drop outliers update cli
Hi @josephmje , I might just make a direct PR there then, but will sync to your changes as applicable. Thanks for the heads up! |
Okay everybody ( @akeshavan, @richford, @arokem, @josephmje, @nrajamani3 )
I've overhauled nearly everything. Feel free to accept whatever elements of this PR that you deem useful and agreeable. I'm very flexible about how we go about this from here.
*Stripped of everything freesurfer (at least for now since it was weighing down the package and is not necessary for dmri preprocessing).
*BIDS bindings still need to be updated (I'm working off of a homemade API-- get_bids_layout)
*metadata from raw .json's are used to automatically generate any arbitrary configuration of acqparameters and index.txt files
*Eddy options have been extended, and no need for a second workflow for non-reverse phase encoded acquisitions. The present workflow works for single phase-encodings as well.
*Regarding qc, dwi (and corresponding bvecs) are forced to RAS+ and isotropic voxels (either 1mm or 2mm) using a slightly more robust set of routine that I've also been implementing in PyNets and NDMG.
*Order of operations has changed-- use ART to remove gross motion artifact first, then run topup/eddy, then remove outliers detected from the susceptibility correction
*I'm working on an API for handling multishell data. Shouldn't take too much longer.
*added gibbs noise ringing and denoising routines (various strategies available) to the end of the workflow.
*Still need to connect to the viewer @akeshavan , and to the eddy qc report. save_json/load_json from nipype were giving me issues -- most likely a version issue -- so I commented out for now.
Please let me know if you have any thoughts, questions, critiques, etc. and I'll respond when I can.
@dPys