Medical image analysis - 25/09/2025

Powerful AI systems using synthetic training data

AI systems for image analysis are only as good as the data on which they are trained. The Göppingen-based start-up MIRA Vision has developed a novel method for generating synthetic, photorealistic images, enabling the efficient creation of large, high-quality training datasets. An intuitive platform also allows researchers to evaluate microscopy images with ease.

Since their discovery in 2004, the grid cells in the brain, which are important for our orientation, have been regarded as a kind of “GPS in the head.” However, scientists at the DKFZ and Heidelberg University Hospital have now shown that grid cells work much more flexibly than previously assumed. In experiments with mice, the researchers found that the cells adapt their activity to different reference points depending on the situation.

Why is blood cancer particularly aggressive in some patients? Researchers at Ulm University Hospital have characterised a mutation in the so-called NOTCH1 gene that significantly influences the prognosis of chronic lymphocytic leukaemia (CLL). Remarkably, this mutation is located in the non-coding region of the gene – an area of DNA long considered less relevant for disease mechanisms.

On the basis of the cooperative relations that have developed over time, Heidelberg University and Harvard University are going to deepen and expand their cooperation. They have signed a Memorandum of Understanding to that effect. As an institutional internationalization project, the state of Baden-Württemberg is providing ten million euros to support this initiative, which is part of the state’s “Global Partnership in Science” activities.

About five percent of lung adenocarcinomas, one of the most common forms of lung cancer, are driven by a faulty fusion of two genes, EML4 and ALK. This fusion results in different variants, and until now, clinicians have treated all patients with these fusions the same way. However, new research led by scientists from the German Cancer Research Center (DKFZ) and Stanford University shows that not all fusion variants behave alike.

With the FeMEA project – Ferroelectric Microelectrodes for Biomedical Applications – Hahn-Schickard is setting a pioneering course in bioelectronics research. The aim of the project is to develop novel microelectrode arrays in which ferroelectric materials are used as functional interfaces in CMOS chips for the first time.