YeastSAM is a model and a framework for yeast cell analysis and mask processing. It provides an intuitive GUI launcher and various tools for generating masks, image registration, outline conversion, and mother-bud pair separation. For detailed documentation, visit YeastSAM.
If you just want to run YeastSAM, you can refer to µSAM to install the framework. We provide custom weights for better yeast cell segmentation that can be downloaded from our GitHub Releases. The custom weights can be used with napari, BAND, CLI, and QuPath.
The dataset used for training and validation can be found at Zenodo: https://zenodo.org/records/17204942, including DIC and the masks.
YeastSAM requires a Conda environment. If you do not already have Conda or Miniconda installed, please install one of them first:
run the following commands in your terminal:
git clone https://github.com/YonghaoZhao722/YeastSAM.git
cd YeastSAM
conda env create -f environment.yml
conda activate yeastsam
pip install .YeastSAM can also be installed using the installer.
- MacOS: YeastSAM.pkg
To start the YeastSAM tools launcher:
conda activate yeastsam
yeastsamThis will open a GUI with four main sections:
- napari: Opens napari viewer for interactive mask generation and editing. You can load our custom weight YeastSAM for better accuracy in budding yeast.
This section is to fix the offset between smFISH image and DIC (as masks are generated from DIC).
- Shift Analyzer: Analyze and detect shifts in your image data
- Apply Registration: Apply image registration corrections
- Mask2Outline: Convert mask files to FISH-Quant compatible outline format
- Mask Editor: You can annotate mask images manually or with CNN & U-Net separation module. Download models at GitHub Releases.
The tools/ directory contains the following utilities:
shift.py: Shift detection and analysisregistration.py: Image registration functionalityMask2Outline.py: Mask to outline conversionmask_editor.py: Advanced mask editing interface
If you are using YeastSAM or the dataset for your research, please cite our paper:
Zhao, Y., Zhu, Z., Yang, S., Li, W. YeastSAM: A Deep Learning Model for Accurate Segmentation of Budding Yeast Cells. bioRxiv 2025.09.17.676679 (2025). https://doi.org/10.1101/2025.09.17.676679
We acknowledge the following reference for inspiring our work:
Archit, A., Freckmann, L., Nair, S. et al. Segment Anything for Microscopy. Nat Methods 22, 579–591 (2025). https://doi.org/10.1038/s41592-024-02580-4
See the full COPYRIGHT notice for details.