The version was pushed to 5.0.0 according to Semantic Versioning due to breaking changes in the API.
JIPipe 5.0.0 introduces a new data viewer and a variety of additional new nodes and further enhancements.
Upgrade notes
- Cellpose: the handling of images was changed, so in some edge cases the algorithm might behave differently
- All expressions: the internal handling of some variables has been redesigned. While now each expression now has access to additional variables, there might be issues if your workflows relied on how specific variables were overwritten (i.e. the order of how the variable set is created)
New in Beta 6
- Graph editor performance was enhanced (design is slightly changed)
- New nodes: K-Means/Fuzzy K-Means/DBSCAN clustering on table rows
- New node: Change 2D/3D ROI properties from table
New in Beta 5
- New file chooser: as newly designed file chooser dialog (Open/save files) was added, as the current one loses features due to Java 21 compatibility. The new file chooser is aware of the current project and has access to the project directories. Existing bookmarks are kept. If you prefer the old file chooser, you can navigate to
Project | Application settings | UI | File chooserand enable theAdvancedLegacyfile chooser. - New node: grow/shrink ROI
- New node: separate touching labels (will also return the borders between the labels as mask)
- Data exporter nodes: the path parameter now has a button to directly choose the output
- Run sets: you can now add nodes to a run set that can be executed from within the project overview. This enables providing users of your workflow with various run options (e.g., prediction vs training)
New in Beta 4
- Selection of default artifacts was improved
- Project overview now allows to update artifacts to new revisions (bugfixes) or switch to (non)-accelerated versions
- New node to annotate OME TIFF files with information about its contained data (number of series, OME XML, …) without reading the whole file
- Fixed broken image montage node (caused by regression)
- Fixed balloon tooltips not disappearing if project is switched
- CLI now allows to overwrite project-wide parameters (key is
/[parameter key]) - Fixed Cellpose nodes only processing one image
New in Beta 3
- Author metadata now carries ORCID information
- JIPipe now requests a set of application-wide authors (users are notified)
- Application-wide authors are automatically written into edited projects
- Support for global parameters was added. The global parameters can be accessed via
_global.[key]from most expressions - Most expressions now have variables that are relevant to the project or node (text annotations, user directories, global variables)
- Project overview was redesigned and upgraded to a viable entry point for users of a workflow (e.g., now compartments can be executed from within the overview)
- The project overview now generates hints regarding improvements for the projects (usage of compartments, setting a license, etc.)
New in Beta 2
- New required update site:
N5 - Basic support for OME ZARR (including OME XML according to how Qupath does it)
- Basic support for zipped OME ZARR
- Enhanced performance for editing larger projects
- Filament processing: new node to copy vertices across Z/C/T
- Cellpose 3 (segmentation and image restoration) is now available. Cellpose 3 is handled separately from Cellpose 2
- Cellpose 2/3: Support for time series and disabling multi-channel (was throwing an error in older JIPipe versions)
- Cellpose 2/3: In the case of multi-channel RGB images, JIPipe will now convert the image to greyscale (was throwing an error in older JIPipe versions)
- Old Hessian node was deprecated (replaced with new Hessian node)
- New FeatureJ Hessian 2D/3D node
- New node: FeatureJ Structure 2D/3D node
- New node: FeatureJ Derivatives 2D/3D node
- Enhanced data trace function
- Data viewers now have a button for data trace
- New nodes to apply ImageJ contrast enhance per slice (conversion to 8/16 bit and in-place)
New data viewer
The included viewer for cached data (double-clicking on an item in “Results”) was fully redesigned and standardized, resolving inconsistencies between different data types and missing information. For example, now users will be able to read the annotations associated with the currently viewed data.
Changes to the image viewer
The custom JIPipe 3D image viewer was removed due to usability, stability, and performance issues. All images are now opened in a 2D viewer. To view images in 3D, you can export the images to the ImageJ 3D viewer or Napari.
Image processing
- JIPipe now provides access to image registration algorithms (based on TurboReg). Expressions are used to allow for more flexibility regarding registration.
- A “Concave Hull” algorithm for ROIs is now available
- A new “ROI mirror” operation was introduced that allows to mirror an image over a line ROI
- Scale 2D now allows to paste the image as-is into the scaled canvas
- New node to split labels that are neighboring, but are not in the same connected component
- Crop 2D now adds information about the crop (for restoration)
- New all-in-one ROI drawer for lines, rectangles, and ovals
- New node to convert a table into point ROIs
- Added FFT Bandpass filter
Filament processing
- Filament vertex colors can now be edited
- Filament vertex and edge metadata fields now can be edited
- New node: filaments to labels
Napari integration
- Napari is now available from the “Tools” menu
- The image viewer can now export to a new Napari instance
User interface
- A “Calculator” panel where expressions can be tested was added
- Node examples can now be loaded again from a node (via it’s Quick Actions button)
Smaller enhancements
- Fast Image Arithmetics now has an IF_ELSE function
- A variety of filter-like parameters were renamed to better indicate how the filtering works (e.g., “Only keep xyz if …” instead of “Filter)
- “ROI” (ImageJ ROIs) was universally renamed to “2D ROI” to better indicate that they are 2D ROI
- Cache-aware UI is now debounced, meaning that UI operations will not slow down data caching anymore
- Graph annotations can now be locked
- Temporary paths (Update cache/runs) are now generated in a different way, avoiding errors regarding long paths
- Moved the “Plugins” button into the “Tools” menu
- Enhancements with the dark mode
- Enhanced design
- The command line interface gained additional features
- JIPipe now automatically detects CUDA-compatible graphics cards
- The status bar now has an additional control for acceleration (CUDA etc.) settings
Bugfixes
- Fixed R and Python path quoting issues on Linux and macOS
- Fixed artifact resolver being unable to find correct artifact
- TrackMate: fixed behavior of Spot penalty parameters
- Connect filament vertices was missing direction variables in some cases
- Some common units were missing
- Fixed issue with “Recent projects” list if multiple instances of JIPipe are running
- Fixed issues with the overlap filter
- “Go to” error is now working again