Deep Learning in Medical Imaging

Many of our projects produce large collections of data. These big data sets stem from population-based studies, interdisciplinary clinical research projects, or from freely available crowd challenges. Using state-of-the-art machine learning techniques such as convolutional neural networks and other deep learning architectures, we explore the wealth of information contained in this clinical data.

 

Preparing your data – Training your AI – Building your solution

We support quantitative medicine with our QuantMed modular software platform. Our goal is to enable more reliable, accurate, and efficient clinical decisions. QuantMed offers support along the way: creating reference training data, training and validating deep learning models, and deploying the results into your quantitative diagnostic software. Improving a model requires experts only to correct its predictions and to update the training data collection. With QuantMed, partners at different institutions can extract medical knowledge together without pooling sensitive patient data. QuantMed nodes at each institution locally extract knowledge into knowledge modules. The QuantMed hub collects the knowledge modules from the nodes, fuses them, and redistributes the update to the nodes. Knowledge modules are safe to share under data protection regulations. They do not contain any patient or institution-specific data, only the required technical information to use them in AI products.

We offer end-to-end support for translational research. Your idea, your data, your envisioned solution – our modular toolchain connects the dots

 

Key features

QuantMed offers extended annotation capabilities for training your data
Systematic generation, accumulation, validation, and utilization of quantifiable medical knowledge
Full integration with MeVisLab for building your software solution

Preparing your data

Fraunhofer MEVIS offers software and services for annotating your data based on SATORI (Segmentation and Annotation Tool for Radiomics and Deep Learning).

 

Key features:

Configuration: easy configuration to your clinical problem and data
Extensions: special annotation, visualization, and analysis tools can easily be integrated according to your needs
Integration: direct connection to deep learning framework so that annotations can directly be used

 

Try out our free annotation tool MEVIS draw.

 

Training your AI

We support efficient model building based on the deep learning framework of your choice. All MEVIS tools are at your disposal, offering data preprocessing, cluster-based training, result validation, and deployment into any product. You only need a web browser to orchestrate your AI developments.

The core of our training infrastructure is RedLeaf, which allows using arbitrary deep learning frameworks and connecting them with MeVisLab. Trained models can be used directly in MeVisLab and can also be exported to other applications.

 

Key features:

Web-based configuration and training monitoring
Data preprocessing and augmentation using MeVisLab
Distributed deep learning
Cluster support
Deployment via MeVisLab or TensorFlowServing

 

Building your solutions

Fraunhofer MEVIS develops solutions for directly integrating them into your products, including:

  • Segmentation
  • Registration
  • Detection
  • Classification
  • Body-part regression
  • Endoscopic video analysis
  • Image reconstruction

 

© Fraunhofer MEVIS
© Fraunhofer MEVIS