ModularImageAnalysis (MIA)

New modules

• CreateClassImage creates an image for multiple object sets, where all pixels within an object set have a user-defined index. This way, class images for deep learning training can be easily generated. For example, all nuclei pixels can be assigned a value of "1", all cytoplasmic pixels a value of "2" and so on. The most recently added pixels are written onto the image, irrespective of what's already there, so object overlap can be handled by selecting the dominant objects last in the list.
• CreateObjectGrid will generate a dense array of objects in 2D that fill an image region. The area of each cell in the grid can be specified. This is useful for modules such as ApplySegmentAnything, where the centroids of this grid can be used as inputs.

New module features

• In AddTracks, track rendering with a limited history (i.e. only showing most recent 10 frames) can fade out.
• The maximum output object area can now be defined in ApplySegmentAnything. This prevents excessively large objects being detected, which can take a long time for large image areas.

Other

• Numeric parameters no longer require any value. This means the value can be removed prior to insertion of dynamically-generated fields.
• General bug fixes.