v0.25

• Added test suite with 17 test cases covering:
• Added GitHub Actions CI/CD workflow for automated testing on Python 3.8-3.12
• Fixed field map intensity shift caused by missing normalization within local mask (PR #6)
  • Added mean subtraction within brain mask to correct intensity shift
  • Added mask default option as None for improved flexibility
• Fixed command-line parsing bug in argument handling
• Improved save flag behavior with proper default values and explicit True/False handling

v0.24

• Critical bugfix
• Included label.json
• Updated NIfTI file description header with some information such as TE, TR, B0

v0.23

• Add seed parameter with default value 42 for reproducibility
• --version fixes
• Voxel size of the reconstruction is correctly separated from the voxel size of the object data
• Fixed bug with how --suffix is used
• --save-field now also generates and saves the local field
• Added necessary dataset_description.json, .bidsignore, and README files to BIDS output

v0.22

• Fix a bug where generated data did not always reflect the desired voxel size due to a bug in qsm_forward.resize

v0.21

• Set --save-phase to true by default

v0.20

• Added --save-phase and --suffix as parameters on qsm-forward command-line entrypoint

v0.19

• Use the original data folder from the QSM challenge 2.0 in-silico head phantom, rather than a modified version of it. This should make it easier for users with access to the data to use this package.

v0.18

• Added simple phantom generation options to main entrypoint

v0.17

• The B0 direction now correctly alters the affine to simulate an oblique acquisition

v0.16

• Improved BIDS compliance: made session and run entities optional; added acq entity, improved argument parsing in CLI, now using derivatives directory for 'derived' types such as masks, segmentations and fieldmaps; also used suffixes correctly including _Chimap, _fieldmap, _dseg and _mask, including the _desc- entities when multiple derivatives are generated that would otherwise clash.
• Removed docs since they were difficult to maintain with minimal effort - the docstrings on the functions are good enough for now and the examples should make it clear how to use the software