KNOSSOS visualizes large 3D (up to multiple terabyte) volume electron microscopic (e.g. Serial Block-Face EM) datasets by displaying slices. It supports skeleton and volume-based annotation modes, which can be extended by plugins written in Python.

It is developed at the Max Planck Institute for Medical Research in Heidelberg, Germany, for Windows, GNU/Linux and OS X. A similar, web-based implementation is being developed at webknossos.info.

Take a look at KNOSSOS’ features to learn more about it:

3D Visualization

3D visualization of image datasets is done by displaying a 2D representation of each side, and allowing you to navigate through these image planes. By dynamically loading only data from the surrounding of the current location, seamless navigation is not limited to datasets that fit into the available RAM but also works with much larger datasets stored in KNOSSOS’ special format on disk.

3D Annotation

KNOSSOS supports two annotation methods—skeletonization as well as 3D segmentation for volume reconstruction. Skeletonization is done by placing and connecting nodes, while KNOSSOS’ segmentation mode allows manual processing of pre-segmented data, and creation of new segmentations from scratch.

These features are already being used at Max Planck Institute for Medical Research (among others), where mice retina was successfully reconstructed.


You can enhance KNOSSOS’ annotation features by writing Python plugins, and we also provide a Python script that helps you to convert your existing image data into a KNOSSOS-readable format.

Open Source & Cross-Platform

KNOSSOS is developed using the Qt5 toolkit, and available on all major platforms. You can help the development of KNOSSOS by submitting bugs and other suggestions at Github’s issue tracker or by contacting us directly.

Getting Started

Download KNOSSOS

KNOSSOS is available for Windows, GNU/Linux and OS X. Since KNOSSOS is an open-source program, we host our releases on Github. Head to our Github repository to download the appropriate version.


A special image format is required to use KNOSSOS. You have the choice to either try our example datasets, or learn how to create your own. Any dataset can be loaded into KNOSSOS by selecting its .conf file in File - Choose Dataset..., and clicking on Use.

Built-in Datasets

There are two datasets already built-in into KNOSSOS: e2006 and ek0563. You can access them by opening the Choose Dataset… window, and selecting that dataset in the table.

Offline Datasets

These are offline datasets that already contain all pre-formatted images: small (54 MB), large (400 MB)

Own datasets

If you have your own image datasets, they will probably need to be converted into KNOSSOS’ format. Head to the Dataset Preparation section to learn how to do so.

Feel free to contact us if you need any assistance in setting up KNOSSOS.

About us

KNOSSOS is developed by a group of students from Heidelberg University and Mannheim University of Applied Sciences, employed at Max Planck Institute for Medical Research.

Ph.D. student

Jörgen Kornfeld

Ph.D. student

Ph.D. student

Fabian Svara

Ph.D. student

Student of Medical Informatics

My-Tien Nguyen

Student of Medical Informatics

Student of Computer Science

Norbert Pfeiler

Student of Computer Science

Student of Computer Science

Michael Pronkin

Student of Computer Science

Student of Molecular and Cellular Biology

Oren Shatz

Student of Molecular and Cellular Biology

Student of Computational Linguistics

Sebastian Spaar

Student of Computational Linguistics

Student of Computer Science

Alex Stepanov

Student of Computer Science

Contact us

If you have any questions or suggestions regarding KNOSSOS, feel free to write us: