QuantEM

An optimized platform of vision transformer-based models for segmentation and analysis of electron microscopy data.

QuantEM is an open-source platform for segmentation and quantitative analysis of electron microscopy (EM) data across imaging modalities, tissues, and species. It provides pretrained models for mitochondria, endoplasmic reticulum, nuclei, and lipid droplets, together with interactive proofreading, guided fine-tuning, and downstream quantitative analysis.

This repository is the code and data resource accompanying the QuantEM manuscript, available at https://www.biorxiv.org/content/10.64898/2026.08.06.743293v1

---

What is here

Directory	Contents
training_and_analysis/	All code behind the manuscript: dataset assembly, tiling, foundation-model pretraining, decoder training, benchmarking, and the immuno-EM analysis.
dataset_directory/	A browsable, filterable directory of the assembled EM image corpus, with a thumbnail for every image and volume. Served as a static site.
quantem_app/	The QuantEM application: segmentation, manual annotation and interactive proofreading, guided fine-tuning, and built-in quantitative analyses, running offline on one machine. Installable as a Python package or as a Windows desktop installer. Under active construction.
quantem-core/	The inference and lightweight-adaptation engine for the eight released models — the shared core the napari plugin builds on.
napari-quantem/	The QuantEM and OmniEM segmentation models packaged as a napari plugin, with widgets for segmenting, fine-tuning, proofreading, measuring and batch runs.
pages/	The landing page for the project's GitHub Pages site, which also hosts the dataset directory.

Data availability

Source EM and segmentation data is available via public repositories, reference the Supplementary Tables with the manuscript for a full list of sources.

Model weights are available at https://huggingface.co/ArrojoeDrigoLab/quantem

Citation

If you use QuantEM — the software, the released model weights, or the dataset directory — please cite:

Acree C, Krystofiak E, Coate K, DelGiorno KE, Winn NCE, Novak SW, Zaganjor E, Magnuson MA, Arrojo e Drigo R. QuantEM: An optimized platform of vision transformer-based models for segmentation and analysis of electron microscopy data. bioRxiv 2026.08.06.743293 (2026). doi: 10.64898/2026.08.06.743293

@article{acree2026quantem,
  title   = {QuantEM: An optimized platform of vision transformer-based models for
             segmentation and analysis of electron microscopy data},
  author  = {Acree, Christopher and Krystofiak, Evan and Coate, Kathryn and
             DelGiorno, Kathleen E. and Winn, Nathan C. E. and Novak, Sammy Weiser and
             Zaganjor, Elma and Magnuson, Mark A. and {Arrojo e Drigo}, Rafael},
  journal = {bioRxiv},
  year    = {2026},
  doi     = {10.64898/2026.08.06.743293},
  url     = {https://www.biorxiv.org/content/10.64898/2026.08.06.743293v1}
}

Machine-readable metadata is in CITATION.cff.