napari-pitcount-cfim

License

BSD 3-Clause

About

This napari plugin was developed in partnership with CFIM (Centre for Microscopy and Image Analysis, Copenhagen University).

The plugin enables image analysis for microscopy, focused on identifying pits and segmenting cells, then generating detailed statistics. It is tailored for using .czi files and integrates well with the napari-czi-reader.

For training the VGG19 2_2 × Random Forest Classifier used in this plugin, visit the pitcount-ml-training repository.

Features

• Detects pits in images using a trained torchvision model.
• Performs cell segmentation via Cellpose (default model: cyto3).
• Calculates and outputs statistics such as:
  • Total cell count
  • Total pit count
  • Percentage of cells containing pits
  • Average number of pits per cell

Usage

Graphical Mode (GUI)

You can launch the plugin in napari with:

napari-pitcount-cfim --dev

or open napari and activate the plugin manually.

Headless Mode (NO GUI)

napari-pitcount-cfim --no-gui 

Run --help to list all options:

napari-pitcount-cfim --no-gui -h

Command-Line Arguments

Notes

• --input-folder must be used with --no-gui.

• --pit-mask-folder must be a valid existing directory.

• Set environment variables are used internally to control behavior.

Requirements

Napari recommends installing napari seperately, as it is not included in this package. You can install it with:

pip install napari[all]

Or you can just

pip install napari-pitcount-cfim[napari]

Known Issues

• The plugin might not support the formats of most model output.
• It's not possible to link masks directly to images in the GUI.
• The default pit model, is a stub and mostly for decoration.