Page MenuHomePhabricator

MITK v2022.04
Updated 626 Days AgoPublic

Version 9 of 9: You are viewing the current published version of this document.

Previous changelog โ€ข Next changelog

NOTE: This is a release changelog. It is composed of a selected short list of highlights since the last release MITK v2021.10 - split into dedicated user and developer sections. See the past four changelogs starting from 2021 Week 48 (Very Early December) for a comprehensive, developer-centric overview of changes.

News for MITK Workbench users

Segmentation

For MITK v2022.04, we focused mainly on segmentation and we are already excited about your feedback on many improvements and a streamlined segmentation workflow. Highlights include:

  • The Segmentation and Multilabel Segmentation Views were finally merged into a single Segmentation View
  • Default label names and distinctive colors for new labels further reduce interuptions of your workflow
  • Segmentation tools are now more aware of the currently active label and labels in general, providing improved intuitive behavior
  • Custom default labels for new segmentations greatly streamline repetitive segmentation tasks
  • Mapping of multi-label segmentations is now supported by Image Registration

Introducing nnU-Net in MITK

We are happy to officially announce the integration of the fantastic nnU-Net for fully automatic segmentation tasks. Currently experimental and only available in our Linux installers, your local nnU-Net Python environment setup can be used in MITK to infer, view and further process segmentation masks seamlessly with other tools. Check out the user guide for more information.

๐Ÿ’” Known issues

  • DICOM file names/paths with non-ASCII characters (e. g. German umlauts) may fail to load on Windows. Two different workarounds are:
    • Rename the file
    • Switch to the UTF-8 locale of Windows 10 2019-03 or later before starting the MITK Workbench
  • The label highlight in the Segmentation View may get out of sync with the actual active label when changing the lock or visibility state of other labels. Workaround: Explicitly select the active label after changing visibility or lock states of ther labels.
  • Registration Evaluator's swipe mode: When using the swipe mode to inspect the registration results, the swipe rendering is always one interaction behind. So it will be rendered with the position you clicked before the current crosshair position. ---

News for developers

ITK v5 upgrade

The biggest change in the backend - besides all the work on segmentation - is the major upgrade of ITK, one of our core dependencies, to version 5. While the upgrade will give us some benefits and access to the latest features and algorithms of ITK for upcoming releases, we focused on the migration of existing code for this release to continuously provide a stable and robust user experience with MITK. When you upgrade your code base to MITK v2022.04, you probably need to adapt some code as well. Scan the previous changelogs for more information and examples on how to migrate existing code.

๐Ÿ›  Third-party dependency changes

NOTE: Here is the list of supported platforms for MITK v2022.04.

The following table shows a complete list of changed third-party dependencies.

DependencyOld versionNew version
Boost1.741.78 beta 1
ITK4.13.35.2.1
VTK9.0.19.1.0
GDCM3.0.83.0.10
OpenCV3.4.83.4.16
Eigen3.2.83.4.0
JsonCpp?1.9.5

๐Ÿ”ฅ API-breaking changes

Besides the inevitable API-breaking changes after the ITK v5 upgrade, most other minor API-breaking changes are summarized here as well. In addition, some legacy classes and properties in the context of Volume Visualization were removed, as listed in the previous changelog.

Previous changelog โ€ข Next changelog

Last Author
kislinsk
Last Edited
Jul 11 2022, 9:10 AM