napari-epyseg 0.0.8

Napari plugin to segment epithelia with EpySeg

napari-epyseg

Plugin to run EpySeg directly in Napari. Can handle temporal data (movie) or single time point (image).

Installation

To install it from Napari, go into the Plugins menu, select Install/Uninstall Plugins.. and look for napari-epyseg in the plugins list.

To install it directly outside of napari, create/reuse and activate a python environement, e.g epyseg_env and install it with pip:

pip install napari-epyseg

[!IMPORTANT] epyseg is compatible until python 3.10 (included), but not for versions of python above. Thus, napari-epyseg is also compatible with python versions until 3.10

Usage

In Napari, go to Plugins>napari-epyseg to start it. It will open an interface in the left part of the main window.

You must select the layer (image or movie, single color channel) on which to run EpySeg. For this, in the Pick an Image parameter, select the corresponding layer (you should open the image/movie independantly of the plugin).

To run EpySeg with default parameters, press directly Segment once you have selected the image. When processing is finished, a new layer called Segmentation will be added in the right panel of the interface. You can save the result with the Save segmentation button that appears on the left part of the interface. Choose where to save the file and the file name with the Segmentation filename parameter, and click the button to save it.

Remark

This plugin was tested on python 3.10, with epyseg version 0.1.52, napari version 0.4.19, tensorflow 2.14

Troubleshooting

• On Linux Ubuntu, python 3.10, epyseg don't run with tensorflow 2.15 and output the error: DNN not found. Downgrading tensorflow to 2.14 worked.
• tifffile versions compability: some versions of tifffile where not compatible with epyseg, but recent versions are now fine. tifffile <= 2021.11.2 are fine and tifffile==2025.1.10 also.

License

This plugin is distributed under the BDS-3 license. If you use this plugin, please cite Epyseg and the plugin with its doi: