Qsiprep

========================

Major additions to the reconstruction workflows! Most notably PyAFQ is available
as a reconstruction workflow. The default atlases included in QSIPrep have been
updated to include subcortical regions if they weren't already present in the
original atlas.

* Add PyAFQ reconstruction workflows 398 Credit: 36000
* Make sure all recon workflows respect omp_nthreads 368
* Add DKI derivatives 371
* Properly transform 4D CNR images from Eddy 393
* Update amico to version 22.4.1 394
* Fix concatenation bug 403 credit: cookpa
* Prevent divide by zero error 405 credit: cookpa
* Critical Fix, use correct transform to get atlases into T1w space 417
* Add resampled atlases back into derivatives 418
* Add connectome2tck exemplar streamlines for mrtrix connectivity workflows 420
* Update the atlases to include subcortical regions 426 [details here](https://github.com/PennLINC/qsiprep-atlas/blob/main/QSIRecon%20atlases.ipynb)

==================================

**WARNING** There is an bug in the connectome pipelines that makes the connectivity
matrices unreliable. Do not use this version for connectome estimation.

Due to persistent difficulties with crashing ODF plots in the reconstruction workflows,
there is now a `--skip-odf-reports` option that will disable the ODF and peak plots
in the html reports. This should only be used once you've run some test workflows
with the reports still enabled, so you know that your ODFs are correctly oriented.

* Make ODF Plots optional (364)
* Bugfix: ABCD gradient data for extrapolation (363)
* Adds `dipy_dki` reconstruction workflow (366)

======================================

**WARNING** There is an bug in the connectome pipelines that makes the connectivity
matrices unreliable. Do not use this version for connectome estimation.

A lot of changes in QSIPrep. The big-picture changes are

1. The build system was redone so a multistage build is used in a
different repository (https://github.com/PennLINC/qsiprep_build).
The container should be about half as big as the last release.
2. The way anatomical masks are handled in reconstruction workflows
has been changed so that FreeSurfer data can be incorporated.
3. FAST-based anatomically-constrained tractography is now deprecated in
QSIPrep. If you're going to use anatomical constraints, they should be
very accurate. The hybrid surface-volume segmentation (HSVS) is
*amazing* and should be considered the default way to use the
MRtrix3/3Tissue workflows. The
[documentation](https://qsiprep.readthedocs.io/en/latest/reconstruction.html)
describes the new built-in workflow names.
4. The reconstruction workflows have been totally refactored. This won't
affect the outputs of the reconstruction workflows, but will affect
anyone who is using intermediate files from the working directory.
The working directories no longer have those unfortunate `..`'s in
their names.
5. FSL is updated to 6.0.5.1!

Since these are a lot of changes, please be vigilant and check your results!
The QSIPrep preprocessing workflows have not changed with this release, but
the dependencies have been upgraded for almost everything.

* Update FSL to 6.0.5.1 (334)
* Move ODF plotting to a cli tool so xvfb is handled more robustly (357)
* Better FreeSurfer license documentation (355)
* Edited libQt5Core.so.5 so it's loadable in singularity on CentOS (336)
* Fixed typo in patch2self (338)
* Inaccurate bids-validator errors were removed (340)
* Bug in `--recon-input` fixed 286
* Correct streamline count is reported in the mrtrix connectivity matrices (330)
* Add option to ingress freesurfer data (287)
* Add Nature Methods citation to dataset_description.json
* Refactor build system (341)
* SHORELine bugfixes (301)
* Bugfix: handle cases where there is only one b=0 (279)

===========================
Change in behavior in Patch2Self:

* Updates Patch2Self with optimal parameters (use OLS instead of ridge)

======================
Bugfixes and documentation

* Updates documentation for containers (270)
* Fixes a bug when reading fieldmap metadata from datalad inputs (271)
* Change incorrect option in the documentation (272)

=====================
Adds a new reconstruction workflow for the NODDI model.

* Adds NODDI reconstruction workflow (257). Thanks cookpa!
* Fixes issue with unequal aspect ratios in q-space plots (266)