Machine Learning

Machine Learning

Machine learning with its capability to recognize complex patterns in big data is currently transforming cancer research. Many groups at DKFZ employ machine learning and deep learning to advance cancer research and treatment. Application areas include imaging and diagnostics, the analysis of high-dimensional omics datasets, the linking of clinical, experimental, imaging and omics data, and surgery and intervention. Methodologic research foci are causality, uncertainty, interpretability and scaling.

Bangert group

Radiotherapy Optimization

Our work revolves around the computational incorporation of physical and numerical models into the treatment optimization process for cancer therapy with radiation.

Proportion of this group's activities taken up by computational research: 80%

Floca group

Software development for Integrated Diagnostic and Therapy

Computational research in the context of uncertainty quantification and hyper parameter optimization of image processing pipelines as well as medical image registration.

Proportion of this group's activities taken up by computational research: 50%

Maier-Hein group

Medical Image Computing

The group develops machine learning algorithms, mathematical modelling approaches and the required research software infrastructure for computational image understanding and large-scale information processing. 

Proportion of this group's activities taken up by computational research: 100%

Lena Maier-Hein group

Computer Assisted Medical Interventions

The vision of the division is to observe everything occurring within and around the surgical treatment process in order to provide context-aware assistance to the physicians.

Proportion of this group's activities taken up by computational research: 90%

Stegle group

Computational Genomics & Systems Genetics

The Stegle team uses statistics and machine learning to analyse and integrate genomic variation datasets. 

Proportion of this group's activities taken up by computational research: 100%

to top