Fine-tuned Cellpose-SAM models for cell and nucleus segmentation in SBF-SEM images of tumor spheroids
-
1.
Swiss Federal Laboratories for Materials Science and Technology
-
2.
ETH Zurich
-
3.
Universitätsklinik Balgrist
-
4.
Center for Microscopy and Image Analysis
-
5.
Scientific Center for Optical and Electron Microscopy
-
6.
Otto-von-Guericke-Universität Magdeburg
-
7.
Federal Institute For Materials Research and Testing
Description
This Zenodo record contains the fine-tuned Cellpose-SAM model weights for cell and nucleus segmentation in serial block-face scanning electron microscopy (SBF-SEM) images of tumor spheroids, as described in:
Bottone et al., 3D Reconstruction of Nanoparticle Distribution in Tumor Spheroids with Volume Electron Microscopy
Two models are provided, targeting cells and nuclei respectively. Both were fine-tuned from the pretrained Cellpose-SAM baseline on manually annotated SBF-SEM images of FaDu head-and-neck carcinoma spheroids resampled to isotropic 100×100×100 nm voxel spacing. Each model folder contains the model weights alongside the training configuration (config.json) and training manifest (training_manifest.json) for full reproducibility.
The models can be loaded and used directly via the sphero-vem library (https://github.com/dv-bt/sphero-vem) or the standard Cellpose API. The annotated training dataset is available at BioImage Archive (https://doi.org/10.6019/S-BIAD3263).