Upgrade to 0.6.0 #126
Upgrade to 0.6.0 #126
Conversation
- Make use of MRtrix3 dwicat command, which means that multiple input DWI series will have their intensities balanced based on their relative b=0 intensities. In addition, if the input DWIs have different voxel grids, they will be conformed onto the same voxel grid prior to running dwifslpreproc; this will be done using rigid-body transformation if possible. - In addition, if contents of the fmap/ directory are to be used, they will be explicitly aligned to the DWI voxel grid for construction of the SE-EPI image to be used as input to dwifslpreproc, avoiding interpolation if possible.
|
Hi. Thanks for all this work @Lestropie! |
|
Unfortunately, there was an issue during processing using the new code. Below is the error message.
As I said yesterday, this can be low priority as the subject that encountered this issue has already been sent for assessment. |
An instance was discovered where FSL eddy fails due to judging the DWI data to not be a shelled acquisition, but the contents that it writes to stdout is some form of binary data rather than the expected error meswsage. With this change, if eddy fails due to an unknwon reason, but it has not yet been attempted with this option included, that second attempt will be made, even though the corresponding error message has not yet been encountered.
In d4b25b3, the version of MRtrix3 was updated to introduce a feature addition to dwicat. However, this involved changing that dependency to the development branch of MRtrix3, which contains changes over and above just the enhancement of dwicat. With this commit, the version of MRtrix3 installed is genuinely 3.0.4, with some additional unrelated hotfixes since that tag, plus the cherry-picked enhancement to dwicat, such that no other major changes to MRtrix3 are introduced.
Had to address #123 due to it precluding processing of data in my own lab.
Results in increment to version
0.6.0as the incorporation ofdwicatwill result in a change in behaviour even for data that previously would have processed to completion.On MRtrix3 I have created a branch off of
3.0.4with a single commit that cherry-picks MRtrix3/mrtrix3#2702 off ofdev. Makes the behaviour of that software dependency a little more reportable.@aaroncapon Do you want to try building a container with this code (and the updated MRtrix3 dependency) and make sure it runs before I merge & tag? CI is completely broken here, repo is highly neglected, so independent verification would be good if not onerous.