DMEA 2022

With around 11,000 trade visitors and 600 exhibitors from 27 countries every year, DMEA is Europe's leading event for the digitization of healthcare.
Through the diverse program with different modules such as congress, training and networking, you can learn more about current developments and products.

Fraunhofer MEVIS at DMEA among others with the following topic:

"SATORI - A flexible AI collaboration toolkit for data annotation, tumor follow-up, and disease quantification"

Annotation and curation of large amounts of data in general and image data in particular is a special challenge in clinical research, but also in scientific and commercial development of AI-based solutions. Moreover, the annotation, training of neural networks, and application of the networks are often distributed across different software solutions, leading to interface problems.

The SATORI annotation platform developed by Fraunhofer MEVIS has functionalities for efficient annotation, segmentation, and image feature quantification. Intuitive interactive contouring tools complement the fully automated segmentation algorithms. The platform is configurable, extensible and can be easily adapted to individual annotation tasks. Radiomics analyses can be performed based on the contours created in SATORI. An integration of the Fraunhofer MEVIS Deep Learning Framework Redleaf enables the training of neural networks. A special feature is the training loop, which allows users to continuously improve the neural networks by adding more training data. With SATORI the whole processing chain takes place in one system. The underlying client-server technology creates optimal conditions for collaborative projects. SATORI is installed on a powerful server and can be used from any end device (PC, tablet, smartphone) via a browser.

SATORI is used by our commercial and scientific partners as an annotation tool for medical data, to support locally distributed in-house AI research and development as well as to conduct clinical trials across sites. SATORI is currently used in more than 15 research projects and collaborations with numerous partners in Germany and Europe. In addition, SATORI is part of the RACOON project and is used in all university hospitals in Germany [https://racoon.network/] for interactive and AI-based automatic segmentation of lung structures and pathologies.

In particular, the "RACOON Lung Analysis Platform" application used in the RACOON project will be presented at the booth. Here, a complete workflow for the segmentation of a lung CT will be shown, including AI-based automatic segmentation, interactive correction tools and 2D/3D visualization. This workflow allows to generate data for research on COVID-19 in the RACOON project. The system is installed at all German university hospitals and has been used for segmentation of well over 2,000 cases so far. In addition, we will show our newly developed training loop, which allows users to improve the AI models by adding their own data for training, and our solution for efficient tumor progression control.

More about the AI Collaboration Toolkit

The experts of the Fraunhofer MEVIS team look forward to welcoming you at the booth of the Fraunhofer-Gesellschaft (hall 2.2, booth E106).