Our aim is to advance scientific understanding by creating practical 3D image analysis solutions for biological structures and materials on multiple scales. We achieve this through the transfer of general-purpose methods to address specific questions. The methods are typically from the fields of image segmentation, image registration, geometry reconstructions, geometric topological analysis, and visualization and quantification. We adapt them for specific image analysis tasks and create software, so that domain scientists can apply our solutions. We develop new methods if available methods are not applicable. A research version of our software is available to scientists, with whom we work closely to evaluate and improve the quality of the image analysis. Mature solutions are also transferred to commercial cooperation partners to make them available as commercial software.

The focus of our recent work has been model-based reconstruction and geometric topological analysis with two main application areas: large-scale microscopy in biology and CT for non-destructive testing. Results include automatic and semi-automatic methods for the reconstruction of actin and microtubules from individual electron tomograms and for the reconstruction of microtubules across multiple serial section electron tomograms. Manual labor has been reduced substantially during analysis, so that the analysis of microtubule organization for large parts of the cell from serial section electron tomography seems now feasible for the first time. For the study of bone healing processes, we developed a method for registration of various 2D and 3D imaging modalities. Furthermore, we developed methods for the geometric topological analysis of pore and material structures from CT of soil and asphalt and methods for the analysis of cracks in concrete. Some of the mature analysis methods for electron tomograms have been made available in the commercial software Amira / Avizo.

Our focus for the next years is to advance analysis in large-scale and high throughput microscopy and tomography. One specific goal is to develop techniques for the analysis of morphology from increasing amounts of image data. Shape models and statistical models for feature extraction and uncertainty quantification will play an important role. Ongoing and planned activities include the reconstruction and analysis of whole microtubule spindles from large-scale serial section electron tomograms and the analysis of further structures in tomograms, like membranes and ribosomes. Building on our results in material analysis, we now work on the analysis and manipulation of material structures to support 3D printing in the field of bioengineering and on the analysis of corrosion processes. Furthermore, we work on the analysis of animal anatomy, like the skeleton of rays and sharks and the shape of snake skulls, and the analysis of core samples to support the study of cold-water coral ecosystem.