Research at the Department of Image Analysis is focused on the development of modern technology and computational methods to automatically process medical multi-modal images with the goal to analyze large longitudinal patient and population studies, to quantify treatment effects, to recognize early structural disease effects, and to identify risk and preserving factors of neurodegenerative diseases. Furthermore, the fast-growing field of precision medicine requires novel, e.g., machine-learning based approaches for the computer-aided diagnosis and prognosis to support treatment decisions tailored to a specific individual at early disease stages. Our mathematical and computer-science oriented research projects are performed in a multidisciplinary environment with a team of population scientists, neuro scientists, clinicians, and MR physicists at the DZNE as well as with our national and international collaborators, e.g. at the Harvard Medical School and Massachusetts Institute of Technology. More details on the research program of the lab can be found here.

We have developed advanced computational methods for the automatic extraction of sensitive preclinical biomarkers from non-invasive MRI, such as, size, thickness and shape of neuro-anatomical structures. Features include cortical thickness maps, the volume of subcortical structures and subnuclei, as well as fiber tracts and white matter lesions estimates. Our methods are distributed via the software package “FreeSurfer” and are employed by thousands of research labs and large cohort imaging studies around the world to perform, for instance, the sensitive quantification of subtle structural changes that occur in early stages of neurodegenerative diseases. The computationally expensive processing of large datasets poses a special challenge in this context and requires parallelization and distribution of the computation (such as GPU, High Performance Computing and Cloud Computing).