Filters

Settings

Now showing1 - 2 of 2
  • Some of the metrics are blocked by your 
    consent settings
    Item type:Project, Access status: without files ,
    Emmy Noether: Computational Soft Material Mechanics Intelligence
    Polymer materials are essential constituents not only of advanced engineered systems but also of everyday products like food, clothes, drugs and consumables, and also all biological organisms consist to a large extent of soft materials. Modeling the mechanical behavior of such materials has proven particularly challenging because they often exhibit complex types of nonlinearity, anisotropy and inelasticity. Descriptive modeling of soft materials therefore still often requires a large amount of expert knowledge. An even greater challenge are predictions how the behavior of soft materials can be tuned by subtle changes of their microstructure and processing. So far such predictions are possible only to a very limited extend and again with considerable expert knowledge. This project aims at leveraging most recent advances in physics-informed machine learning to automate both the descriptive and predictive modeling of soft materials to an unprecedented degree. Thereby, it will create a powerful basis for the future virtual design of soft materials. Moreover, by automating the pipeline from probing to describing to predicting the mechanical behavior of soft materials, the methods developed in in this project will enable robots to examine and handle also unknown and complex soft materials with an unprecedented degree of autonomy. This can be transformative for biomedical and biomechanical robotics, where, for example, surgical robots are facing in each patient or individual a different behavior of soft tissue and have to learn to probe and handle it. To showcase such benefits for medical robotics, this project will not only develop a toolbox of machine-learning based methods for describing and predicting the mechanical behavior of soft materials in a largely automated way but it will also illustrate their application in a simple robot demonstrator in the laboratory.
      36
  • Some of the metrics are blocked by your 
    consent settings
    Item type:Project, Access status: without files ,
    A 3D DCE-US Interconnected Voxel Analysis Framework for Bedside Characterization of Tumor Vascular Properties in Liver Malignancies
    Liver cancer results in a significant number of deaths worldwide. Characterization of patient-specific tumor vasculature and perfusion at bedside can enable better, individualized treatments. We propose a reconstruction method based on a two-compartment contrast flow field model to characterize vascular properties through the analysis of 3D dynamic contrast-enhanced ultrasound measurements. Our inversion method determines patient-specific parameter functions by minimizing the mismatch of simulated and measured contrast concentrations. The analysis is performed on synthetic as well as experimental animal data. To assess clinical relevance, we investigate reconstruction of physiologically interpretable quantities such as arterial blood flow and perfusion.
      74