Cellpose – practical

Trainer: Damian Edward Dalle Nogare

Overview: This section covers Cellpose, a generalist, deep learning-based segmentation method, which can precisely segment cells from a wide range of image types and does not require model retraining or parameter adjustments (https://doi.org/10.1038/s41592-020-01018-x).

Learning outcomes:

After this section you should be able to:

Identify Cellpose as a cellular segmentation algorithm, its potentials and limitations
Perform image analysis with Cellpose

Materials:

Cellpose presentation
Cellpose notes
Cellpose notebook (practical); please use the Linux command-line to download –> wget https://ftp.ebi.ac.uk/pub/training/2024/Microscopy_data_analysis_machine_learning_and_the_BioImage_Archive_2024/Day5/cellpose_notebook.ipynb
Practical data

Further information:

Cellpose main site: https://www.cellpose.org/
Cellpose github: https://github.com/MouseLand/cellpose
Cellpose API reference: https://cellpose.readthedocs.io/en/latest/
Fiji-cellpose bridge: https://github.com/BIOP/ijl-utilities-wrappers

Installation procedure:

conda create -n cellpose
conda activate cellpose
conda install pytorch==1.12.0 cudatoolkit=11.3 -c pytorch
pip install cellpose[gui]==2.3.2 (latest version 2)
pip install cellpose[gui] (version 3)
if there is an error also: pip install pyqt5

Microscopy data analysis: machine learning and the BioImage Archive

Cellpose – practical