MRtrix3Tissue v1.0.0
·
360 commits
to master
since this release
This is the first MRtrix3Tissue production release. This release introduces the new 2019 version of the dwi2response dhollander algorithm, but also brings in a very large amount (~ 1 year) of MRtrix3 developments. Several major changes end users should be aware of include:
- the
dwi2response dhollanderalgorithm has been improved (new 2019 version), mostly impacting the single-fibre WM response function estimation. This should generally allow for a better subsequentdwi2fodmulti-tissue (2-tissue, 3-tissue) fit, for both multi-shell as well as single-shell data. The new algorithm is also faster than the old one. average_responsehas been renamed toresponsemean, with different default behaviour- the old
dwinormaliseanddwiintensitynormfor intensity normalisation in single-tissue pipelines have been recombined intodwinormalise(with algorithm selection). Note thatdwinormaliseis only recommended still for single-tissue pipelines. For multi-tissue (2-tissue, 3-tissue) pipelines,mtnormaliseis recommended instead, to be performed immediately followingdwi2fod. dwi2tensorchanged its default fitting strategy from ordinary least-squares (OLS) followed by iterative weighted least-squares (IWLS), to weighted least-squares (WLS) followed by IWLSmrresize,mrpadandmrcrophave been removed and recombined into the newmrgridforeachshows progress differently, and has several new options; though basic usage remains the same (i.e. backward compatible)
Due to the amount and impact of the changes (which go beyond the aforementioned list), it's generally not recommended to switch between older versions of MRtrix3Tissue or MRtrix3 and this new version during your ongoing projects (experiments, studies, ...); i.e. either stick with your current version to finish up ongoing projects, or alternatively re-run most processing consistently using a "clean" installation of this new version.