Cell Segmentation with Cellpose

Run Cellpose (https://www.cellpose.org/) segmentation on the input image.

This node allows to apply one of the pretrained models ('nuclei' or 'cyto') on the input image. For documentation of the possible settings, please refer to:
%%00009https://cellpose.readthedocs.io/en/stable/settings.html

The component node defines a conda environment with CPU-only inference, and has been tested on Windows and Mac OSX.

DISCLAIMER: In order to use this component and its Python script adopting the KNIPImage package, you need to manually install the "KNIME Image Processing - Python Extensions" available at kni.me/e/CjujPH_iaJn-CwNl

The author of this component node is not affiliated with the authors of Cellpose, nor with KNIME. This node is meant as an illustrative example how to use neural network-based image segmentation in KNIME. Please refer to the original Cellpose paper for information about Cellpose, and to the image.sc forum for questions.

The Cellpose project is Copyright © 2020 Howard Hughes Medical Institute by its authors and released under BSD 3-Clause "New" or "Revised" License (github.com/MouseLand/cellpose/blob/master/LICENSE).

Reference:
Stringer, C., Wang, T., Michaelos, M. et al. Cellpose: a generalist algorithm for cellular segmentation. Nat Methods 18, 100–106 (2021). https://doi.org/10.1038/s41592-020-01018-x

Options

Resample: If selected, will create smoother masks when the cells are large; if deselected, will find more masks when the cells are small.
Input Image Column: Input for segmentation
Expected diameter (pixels): Expected diameter of a single cell, in pixel units
Minimum size (pixels): Minimum size of a cell, in pixel units
Model type: Choice of model to apply (pre-trained versatile models 'nuclei' or 'cyto')
Output column name: Name of the result (segmentation) column

Cell Segmentation with Cellpose

Options

Input Ports

Output Ports

Nodes

Extensions

Links

Download