Skip to content

Version 2.0.0

Choose a tag to compare

View all tags
@joeranbosma joeranbosma released this 31 Aug 09:37
· 27 commits to main since this release
v2.0.0
285deef
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Verified
Learn about vigilant mode.

Major updates:

  • Archives are now converted case-by-case, instead of performing each phase after the other (i.e. collecting metadata for all cases, then converting images).
  • Logging has improved and now offers three levels of verbosity: 1) show full case log when something goes wrong (default), 2) show full log for all cases, and 0) off.
  • Improved usability of command-line interface: both DICOM → MHA and MHA → nnU-Net conversions are now available, as well as the generation of the conversion settings.
  • Conversion of dynamic contrast-enhanced (DCE) scans is now supported (but experimental, feedback is welcome).

Updates that may require you to update your existing scripts:

  • Some parameter names have changed (e.g. input_path to input_dir in MHA2nnUNetConverter). Please refer to the README or source code for the new values.
  • Conversion settings for DICOM → MHA now require the patient_id and study_id for each item.

Minor updates:

  • Multiprocessing is now optional. Set num_threads=1 to turn off.
  • Improved creation of dataset.json for nnUNet raw data archive.
  • Any DICOM tag can now be used to match cases to their target MHA file.
  • The way values from DICOM tags are matched with the provided values is now configurable. See Dicom2MHAConverter's docstring for details.

Improved testing:

  • Test DICOM → MHA conversion from both Python and the command line interface
  • Test conversion of DCE scans
Assets 2
Loading