CONVEA: AI-Based Image Registration for Universal Clinical Use

CONVEA is a spin-off project of the Medical Deep Learning group that translates the ConvexAdam image registration technology developed at the institute from research into clinical practice.

Image registration is a core technology in medical image analysis: it establishes a spatial correspondence between scans acquired at different points in time, using different imaging modalities, or from different patients. This enables precise comparison of anatomical structures and makes changes visible. Registration therefore provides an important foundation for numerous applications, including longitudinal monitoring, treatment planning, motion analysis, and the transfer of image information between CT, MRI, and other modalities.

 

CONVEA combines medical foundation models with optimisation-based registration and automatic self-configuration. The aim is to create a robust and scalable technology that can be used universally across imaging modalities, anatomical regions, and clinical use cases. Funded through the EXIST Research Transfer programme, CONVEA is developing its prototypes together with clinical partners towards initial pilot applications.

EXIST Research Transfer funding (2026–2027): 873,173 €

Selected publication:

Siebert, H., Großböhmer, C., Hansen, L. & Heinrich, M. P. Convexadam: Self-configuring dual-optimization-based 3d multitask medical image registration. IEEE Transactions on Medical Imaging, 2024.

Project team:

Prof. Dr. Mattias Heinrich
Dr. Sabrina Reimers-Kipping
M. Sc. Christoph Großbröhmer
M. Sc. Wiebke Heyer
B. Sc. Juri Rohde