Skip to content



steinbock is a framework for multi-channel image processing

The steinbock framework comprises the following components:

  • The steinbock Python package with the integrated steinbock command-line interface (CLI)
  • The steinbock Docker container interactively exposing the steinbock command-line interface, with supported third-party software (e.g. Ilastik, CellProfiler) pre-installed

Modes of usage

steinbock can be used interactively as well as programmatically from within Python scripts.


At its core, steinbock provides the following functionality:

  • Image preprocessing, including utilities for tiling/stitching images
  • Pixel classification, to enable pixel classification-based image segmentation
  • Image segmentation, to identify objects (e.g. cells or other regions of interest)
  • Object measurement, to extract single-cell data, cell neighbors, etc.
  • Data export, to facilitate downstream data analysis

While all steinbock functionality can be used in a modular fashion, the framework was designed for - and explicitly supports - the following image segmentation workflows:

  • [Random forest-based object segmentation] Zanotelli et al. ImcSegmentationPipeline: A pixel classification-based multiplexed image segmentation pipeline. Zenodo, 2017. DOI: 10.5281/zenodo.3841961.
  • [Deep learning-based cell segmentation] Greenwald et al. Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nature Biotechnology, 2021. DOI: 10.1038/s41587-021-01094-0.

The steinbock framework is extensible and support for further workflows may be added in the future. If you are missing support for a workflow, please consider filing an issue on GitHub.




Issue tracker:

Citing steinbock

Please cite the following paper when using steinbock in your work:


Windhager J, Bodenmiller B, Eling N (2021). An end-to-end workflow for multiplexed image processing and analysis. bioRxiv. doi:

  author = {Windhager, Jonas and Bodenmiller, Bernd and Eling, Nils},
  title = {An end-to-end workflow for multiplexed image processing and analysis},
  year = {2021},
  doi = {10.1101/2021.11.12.468357},
  URL = {},
  journal = {bioRxiv}