Adapts to Your Research
Any analysis method, research tool or service can be containerized and added to the platform as an extension. Kaapana adapts to your research question rather than the other way around.
KAAPANA
An open-source toolkit for building medical imaging platforms.
Curate, process and analyze medical images at scale; on your own infrastructure, across institutions, without patient data ever leaving the site.
About Kaapana
Kaapanafrom the Hawaiian word kaʻāpana, meaning "distributor" or "part"is an open-source toolkit for building state-of-the-art platforms in the field of medical data analysis, with a focus on radiological and radiotherapeutic imaging.
It's a collection of specialized tools and services, not a fixed application. It enables the creation of platforms tailored to specific use cases: large-scale medical image analysis, collaborative research, multi-center studies and more. It runs on your own hardware or in the cloud, from a single virtual machine to a full Kubernetes cluster.
- Open source: AGPL-3.0
- Developed at DKFZ since 2020
- Runs on your infrastructure
- Installed in hospitals and research centers across Germany and the EU
- Research software – not a certified medical device
Open Source
Following established standards and adopting widely used open technologies, Kaapana is highly compatible with existing clinical IT infrastructure (PACS, object stores, etc.) while being secure, modular and easy to extend.
Built for Collaboration
From federated analysis where data never leaves the site to multi-center studies on pooled cohorts – project-based data governance keeps every study, team and dataset cleanly separated on one platform.
Features
From ingestion and curation to processing, analysis and review – Kaapana brings the core capabilities for medical imaging research together in one platform, so you can focus on your scientific work rather than assembling and maintaining the underlying infrastructure. And where the built-in tools end, Kaapana is built to be extended.
Data Curation
Filter, tag and visualize DICOM data by metadata in the datasets view to build study cohorts. Backed by an integrated PACS, metadata index and object store.
Out-of-the-box Methods
Run state-of-the-art methods like TotalSegmentator and nnU-Net on whole datasets with a few clicks, or train and fine-tune your own models on local data.
Digital Pathology
Convert whole-slide images to DICOM, review them in the SLIM Viewer or OHIF, and browse large pathology datasets with automatically generated thumbnails.
App Store & Marketplace
Install algorithms, workflows and applications as extensions with one click, and share your own methods with other Kaapana instances.
Federated Analysis
Train models and run analyses across sites without moving patient data – only models and aggregated results leave the institution.
Interactive Workspaces
Analyze results in integrated JupyterLab and RStudio environments and work on collaborative spreadsheets and reports with Collabora, without data leaving the platform.
Desktop Streaming
Work with full desktop applications like MITK Workbench or 3D Slicer right in the browser, next to your data, or bring any desktop app you need.
See It in Action
A quick look at the platform – each clip shows a different functionality in Kaapana.
Interactive Data Curation
Explore, filter and tag DICOM data by metadata in the datasets view. Carve out study cohorts interactively instead of writing queries.
Workflows
Trigger processing on whole datasets with a few clicks, from segmentation to radiomics. Workflows run reproducibly in the background and deliver results back to the platform.
Extensions
Install new algorithms, workflows and applications with one click from the extension marketplace, and package your own methods so other sites can do the same.
Project Separation
Run many studies and teams on one platform. Data, workflows and access are cleanly isolated per project, every user only sees what belongs to their study.
Desktop Streaming
Open full desktop applications like MITK Workbench or 3D Slicer directly in the browser. Access data where it resides on the platform.
Getting Started
From bare server to running platform in three steps.
Provision a Server
A single Ubuntu or AlmaLinux machine with 8+ cores, 64 GB RAM and 200 GB storage is enough to start.
Install & Deploy
One script sets up Kubernetes and deploys the platform. Use our prebuilt containers or build them yourself from source.
Log In & Explore
Open the web interface in your browser, send DICOM data to the platform, install extensions from the marketplace and run your first workflow.
Built on Established Technologies
Kaapana relies on widely adopted technologies, making it stable and easy to integrate into existing infrastructures.
Kubernetes
Every Kaapana platform runs on a Kubernetes cluster. All services and extensions are deployed, scaled and monitored there, from a single workstation up to multi-node data-center installations.
kubernetes.io ↗
OpenSearch
Kaapana indexes the metadata of every DICOM image in OpenSearch. This powers the gallery-style dataset browser with full-text search, filtering and cohort building across the whole archive.
opensearch.org ↗
Apache Airflow
Airflow is Kaapana’s workflow engine. Processing pipelines are defined as DAGs that orchestrate containerized steps, from anonymization and segmentation to model training and federated analysis.
airflow.apache.org ↗
dcm4che
The dcm4chee archive is Kaapana’s internal PACS. Incoming DICOM data is received and stored there, then indexed and made available to workflows and viewers on the platform.
dcm4che.org ↗
OHIF Viewer
Kaapana integrates the OHIF viewer to display DICOM images and AI results such as segmentations directly in the browser – no local viewer installation needed.
ohif.org ↗
Keycloak
Keycloak provides single sign-on for the entire platform. Users, roles and project-based access to all components are managed in one place and can be connected to institutional identity providers.
keycloak.org ↗
Docker
Every Kaapana component and processing algorithm is packaged as a container image, so the same workflow runs reproducibly on any site – a key building block for multi-center studies.
docker.com ↗
Projects
Platforms built with Kaapana power medical image analysis in a wide range of projects – a selection:
DKFZ wDB
Wissenschaftliche Datenbank – imaging research database at DKFZ
NCT IDAI
Image Data and Analysis Infrastructure for NCT
NCT DICOM Service
Supports the pathology data management infrastructure of oneNCT
Team
We are a team of researchers, students and software developers working at the Division of Medical Image Computing at the German Cancer Research Center. Reach out and get to know us!
Kaapana Leads
Kaapana Devs
Supporting Devs
MIC Board
Alumni
Publications & Citation
A collection of publications related to Kaapana:
If you want to reference Kaapana, please cite this publication.
Kaapana: A Comprehensive Open-Source Platform for Integrating AI in Medical Imaging Research Environments
Akünal Ü, Bujotzek M, Denner S, Hamm B, Kades K, Schader P, Scherer J, Nolden M, Neher P, Floca R, & Maier-Hein K. (2025). Kaapana: A Comprehensive Open-Source Platform for Integrating AI in Medical Imaging Research Environments.
DOI: 10.48550/arXiv.2512.09644Joint Imaging Platform for Federated Clinical Data Analytics
DOI: 10.1200/CCI.20.00045Show full citation
Scherer, J., Nolden, M., Kleesiek, J., Metzger, J., Kades, K., Schneider, V., Bach, M., Sedlaczek, O., Bucher, A. M., Vogl, T. J., Grünwald, F., Kühn, J.-P., Hoffmann, R.-T., Kotzerke, J., Bethge, O., Schimmöller, L., Antoch, G., Müller, H.-W., Daul, A., … Maier-Hein, K. (2020). Joint Imaging Platform for Federated Clinical Data Analytics. In JCO Clinical Cancer Informatics (Issue 4, pp. 1027–1038). American Society of Clinical Oncology (ASCO).
Efficient DICOM Image Tagging and Cohort Curation Within Kaapana
DOI: 10.1007/978-3-658-36932-3_59Show full citation
Kades, K., Scherer, J., Scholtyssek, J., Penzkofer, T., Nolden, M., & Maier-Hein, K. (2022). Efficient DICOM Image Tagging and Cohort Curation Within Kaapana. In Informatik aktuell (pp. 279–284). Springer Fachmedien Wiesbaden.
Real-world federated learning in radiology: hurdles to overcome and benefits to gain
DOI: 10.1093/jamia/ocae259Show full citation
Bujotzek MR, Akünal Ü, Denner S, Neher P, Zenk M, Frodl E, Jaiswal A, Kim M, Krekiehn NR, Nickel M, Ruppel R, Both M, Döllinger F, Opitz M, Persigehl T, Kleesiek J, Penzkofer T, Maier-Hein K, Bucher A, Braren R. Real-world federated learning in radiology: hurdles to overcome and benefits to gain. J Am Med Inform Assoc. 2025 Jan 1;32(1):193-205.
DICOM Whole Slide Imaging for Computational Pathology Research in Kaapana and the Joint Imaging Platform
DOI: 10.1007/978-3-658-36932-3_58Show full citation
Fischer, M., Schader, P., Braren, R., Götz, M., Muckenhuber, A., Weichert, W., Schüffler, P., Kleesiek, J., Scherer, J., Kades, K., Maier-Hein, K., & Nolden, M. (2022). DICOM Whole Slide Imaging for Computational Pathology Research in Kaapana and the Joint Imaging Platform. In Informatik aktuell (pp. 273–278). Springer Fachmedien Wiesbaden.
Towards Real-World Federated Learning in Medical Image Analysis Using Kaapana
DOI: 10.1007/978-3-031-18523-6_13Show full citation
Kades, K., Scherer, J., Zenk, M., Kempf, M., Maier-Hein, K. (2022). Towards Real-World Federated Learning in Medical Image Analysis Using Kaapana. In: , et al. Distributed, Collaborative, and Federated Learning, and Affordable AI and Healthcare for Resource Diverse Global Health. DeCaF FAIR 2022 2022. Lecture Notes in Computer Science, vol 13573. Springer, Cham.
Get in Touch
Questions, ideas or feedback? Reach the Kaapana team through any of these channels.