Fraunhofer MEVIS @ ECR 2025

The European Congress of Radiology (ECR) Vienna of the European Society of Radiology (ESR) for Radiology is a forward-looking, dynamic and service-oriented congress and trade exhibition, known as one of the most innovative meetings of the scientific community and embedded in a unique and inspiring ambience. Within the framework of the trade fair, around 300 exhibitors will present their latest products and services. The European Congress of Radiology in Vienna is a leading medical meeting place for radiology in Europe.

 

This year Fraunhofer MEVIS contributes the following topics:

 

Moderation / Chairperson's Introduction

Bianca Lassen-Schmidt

Interoperability Session

"AI-SC 8 - AI tools supporting radiology reporting process Categories"

Artificial Intelligence, Imaging Informatics, Management/Leadership, Professional Issues ETC
Level: LEVEL II+III
Date: February 28, 2025 | 12:15 - 13:15 CET CME
Credits: 1 Green Friday

 

Oral Presentation

Sven Kuckertz

"Enhancing speed and precision of lesion tracking in follow-up lung CT using deep-learning-based registration"

Research Presentation Session: Oncologic Imaging
RPS 2216 - Structured reporting, radiomics and deep learning

(Authors: S. Kuckertz, S. Heldmann, F. Peisen, J. H. Moltz)

Date/Time: Sunday, March 2,  08:00 – 09:00 CET (UTC+1)

Room: ACV - Research Stage 3

 

Chiara Tappermann

Poster

"Advancing adenomyosis detection through deep learning-assisted uterus segmentation in MRI"

ACV Level -2 in EPOS area

February 28 to March 2 

 

Pixel Pandemonium:

(Come and find us at the EXPO X1, Level -2 in the AIX area.)

The Pixel Pandemonium at ECR 2025 is the perfect platform to explore up-and-coming AI and ML tools for medical imaging. This program offers a unique opportunity to see innovative technology firsthand and get a hands-on experience!
Discover the latest AI/ML technology in the medical imaging sphere. We want to see what is coming and what might have an impact in the next couple of years!
The Pixel Pandemonium Exhibition is endorsed by the Medical Image Computing and Computer Assisted Intervention Society (MICCAI).
Demos in the exhibition include more than 15 software tools showcasing cutting-edge AI technology.

 

Horst Hahn 

Co-Organizer
 

Wednesday & Thursday:

Johanna Brosig

"Mitral valve analysis and AR surgery planning"

Our demo facilitates automatic mitral valve analysis and minimally invasive surgery planning. We generate polygon meshes from nnUNet-based medical image segmentation masks, enabling parameterization of the mitral valve and simulation of mitral valve repair options using position-based dynamics. Additionally, we show that by visualizing the generated left heart models using augmented reality, this approach can be applied to intervention planning and patient education. It addresses the challenge of patient-individual heart assessment in valvular heart disease, which traditionally requires manual interaction and is time-consuming. This method improves efficiency and supports minimally invasive surgery planning.

Friday & Saturday:

Niklas Agethen

"AI based fiber tracking"

This deep learning-based method enables fully automated reconstruction of white matter tracts from raw diffusion MRI data. By bypassing complex diffusion models, it facilitates more efficient fiber pathway analysis, particularly in brain tumor cases. The demo showcases its high reconstruction quality for key tracts (e.g., arcuate fascicle, corticospinal tract, optic radiation) alongside a comparison with conventional probabilistic methods. It addresses the challenge of time-consuming, expertise-dependent fiber reconstruction, improving efficiency and reproducibility in intervention planning.

Kai Geißler

"Uncertainty Guided Training Loop"

Our uncertainty-driven training loop integrates model training directly into the annotation process, allowing experts to refine AI segmentation models iteratively. By quantifying model uncertainty, it guides annotators to the most impactful areas for correction, reducing the need for full dataset re-labeling. It addresses the challenge of inefficient annotation workflows in AI development by focusing expert effort where it matters most, improving training efficiency and accelerating model refinement.

Hannah Strohm

"The AI based assisstant for liver lesions"

ProMedCEUS offers an AI-driven approach for automatically classifying washout patterns in contrast-enhanced ultrasound (CEUS) scans. Given three characteristic contrast sequences and segmented lesion data, the system predicts washout categories with an uncertainty estimate and visual feedback. It addresses the challenge of CEUS interpretation, which requires expertise and hinders widespread clinical adoption, by providing an automated, standardized tool for more reliable liver lesion diagnosis.