Analysis pipelines

Cell Segmentation and shape measurement (ImageJ + Cellpose)

By Kai TongUpdated 2026-10-09

Segments individual cells in images of snowflake yeast clusters crushed into a single cell layer, then measures each cell’s size and shape (area, major/minor axis, aspect ratio, circularity, solidity) and, optionally, fluorescence intensity. Used for the cell aspect ratio and cell volume protocols.

How it runs: (1) an ImageJ macro prepares the images; (2) Cellpose (v2.2.2, built-in “cyto” model, cell diameter 80 px) finds the cells; a GPU is strongly recommended, whether a local GPU, a computing cluster (an example Slurm job script is included) or Google Colab; (3) an ImageJ macro filters the results; (4) optional manual correction in the Cellpose GUI; (5) an ImageJ macro measures each cell and saves a CSV per image.

Needs: Fiji/ImageJ with the BioVoxxel, PTBIOP and LOCI update sites, and Cellpose 2.2.2 (an environment file recreating the exact setup is included). Input: .nd2 or other ImageJ-readable images with a cell-wall stain (e.g. Calcofluor White) or brightfield channel. Step-by-step instructions are in README.rtf. Needs: Fiji/ImageJ with the BioVoxxel, PTBIOP and LOCI update sites, and Cellpose 2.2.2. The included environment file recreates Kai’s exact Cellpose setup. Input: .nd2 or other ImageJ-readable images with a cell-wall stain (e.g. Calcofluor White) or brightfield channel. Full step-by-step instructions are in README.rtf.

Download (.zip, 22 KB)