Version 1.2.7

• ExtractObjectEdges now accepts object and parent object measurements as metrics for the edge width
• FilterMeasurementsByExtremes can now retain/remove N objects with the largest/smallest measurements, rather than a single object
• Added ClearOverlay module to remove any overlay components from an image
• Added MIA_RunWorkflow macro to run a workflow directly from an ImageJ macro without the need for MIA to be open
• Other general bug fixes

Version 1.2.6

• Added "Random (vibrant)" colormap for overlays
• Bugfix for splash screen on MacOS
• Other general bugfixes

Version 1.2.5

• AddLabels can now use positions from object measurements
• Added MorphoLibJ's surface area and sphericity measurements to output of MeasureObjectShape
• Added measurement outputs to RealteOneToOne
• Updated XLSX exporter dependencies so errors and warning shouldn't occur

Version 1.2.4

• Image registration modules now work with images of different resolutions. Both images are padded to the maximum width/height of either image.

Version 1.2.3

• Added bit depth as a measurement for MeasureImageDimensions
• Overlays in ManuallyIdentifyObjects now only appear in their Z-slice
• Added option to MaskObjects to remove objects left with no volume following masking
• Added alternative thresholding option to RelateManyToMany to allow linking as long as one object has a minimum overlap.

Version 1.2.2

• Added in ellipsoid fitting using a couple of BoneJ functions
• Added variable patch size for ApplyDeepImageJ
• General optimisations and bug fixes

Version 1.2.1

• Added ApplyDeepImageJModel module, which runs any DeepImageJ currently installed in Fiji on a provided image (requires DeepImageJ update site to be enabled).
• Added DistanceBands module, which creates concentric band objects around objects in a binarised image
• Added GrowObjects module, which simplifies the application of segmenting cells and cytoplasm. This can perform distance- or intensity-based watershed segmentation of a region from provided marker objects. The output objects are automatically related to their marker object.
• Added more automated tests, primarily for image processing modules.
• SaveObjectsAsROIs now includes additional options for saving objects as individual ROI files or on a per-slice basis.
• Generalised image rendering for improved compatibility with other projects that use MIA as a library.
• General bug fixes and improvements.
• Updated publications on https://mianalysis.github.io/mia/publications site

)
Example image generated using DeepImageJ integration (model is Neuron Segmentation in EM (Membrane Prediction) by Constantin Pape)

Version 1.2.0

• Added headless mode to allow workflows to be run without the need for opening a graphical instance of Fiji. This enables running MIA on terminal-only systems such as compute clusters. For more information see the guide at https://mianalysis.github.io/mia/guides/runningheadless.html
• The "MIA (headless)" menu option can also be used to run existing workflows with a minimal graphical interface and provides a method for launching workflows from ImageJ macros
• Added modules to load and save objects to ImageJ .roi and .zip (zipped ROI) files
• Included StarDistDetection module from mia-deeplearning repository. This module will only appear if StarDist update site is enabled
• Projection axis cannow be specified in ProjectImage. This module also switched to primarily using ImgLib2. ProjectImage also outputs average projections as 32-bit
• Updated handling of modules with incompatible dependencies. Now such modules aren't flagged at launch of MIA; instead, inactive modules are listed in both "Unavailable modules" section of "Add module" menu and via "Help > Unavailable modules" menu. Missing modules are still flagged when loading workflows that call for them
• Minor website formatting updates and removal of "html" from URLs
• Added many "MethodSource" tests for image processing modules. These are needed in the transition of these modules to ImgLib2
• General bugfixes

Version 1.1.1

• Added modules to measure Gini coefficient on whole images (or masked images) and on an object-by-object basis. The Gini coefficient measures the inequality between pixel intensities.
• Added CI{} and CD{} calculation expressions, which output results in integer and double precision formats, respectively. The existing C{} version still exists and will output as a double.
• Added Preference option to specify time elapsed before image loader will generate a BFMEMO file (default 3 seconds)

Version 1.1.0

• Introduces variable parameters. Numeric and text parameters can be injected with expressions that will be evaluated at runtime. With these, it's possible to use values in the workspace as parameters. Currently, the following expressions are supported:

  Type	Expression	Example
  Metadata	Me{[metadata name]}	Me{Filepath}
  Image measurements	Im{[image name|measurement name]}	Im{RedChannel|DIMENSIONS // WIDTH}
  Object measurement statistics	Os{[objects name|measurement name|statistic]}	Os{Nuclei|SHAPE // N_VOXELS|MEAN}
  Object counts	Oc{[objects name]}	Oc{Nuclei}

• Added ability to save and load selections while running ManuallyIdentifyObjects. This should help prevent against total loss of manually-selected regions. Selections are saved to ImageJ ROI .zip file by going to File > Save objects and can be loaded back in by going to File > Load objects.

• Second image in ImageCalculator can be weighted differently for average pixel calculations. This allows one channel to appear stronger when displayed as composite.

• SetDisplayRange can now use an external image to determine intensity ranges (this can provide a speed increase when using "Precise" method)

• Enabled theme selection by going to "Edit > Preferences" from "Editing View"

• Improved Image class abstraction so future inclusion of other image storage methods (e.g. ImgLib2) should be easier.

• Renamed module "WekaProbabilityMaps" to "WekaPixelClassification"