Most of our research is done in the context of real-world usage scenarios. The following lists some of the areas in which our team has experience in researching both application specific algorithms as well as support services and infrastructure.
Simulation of Tumor Thermal Ablation
Laser Interstitial Thermo Therapy (LITT) is an established surgical method. One of the main difficulties of using LITT is the limited capability of the available mathematical models and surgical planning systems to accurately predict the induced damage. Our model simulates the light distribution using the diffusion approximation of the transport theory instead of using the classical approach of Monte Carlo simulation. Furthermore, the model simulates the heat distribution using the bioheat equations including the effect of micro-perfusion in the tissue and the dependency of thermal and optical properties on the temperature and the injury. The injury is estimated using the damage integral. In-vitro validation of the simulation show deviation of 5% on the x-axis till 20% in the y-direction (the scalpel cut was performed in this direction).
Interfacing HPC and Life Science Research Collaborations
Information technology (IT) infrastructures for research collaborations and virtual research businesses are essential elements of science in the field of medicine. The sustainability of the infrastructure depends on how well it integrates into the scientific process and builds on existing skills and knowledge sets. This objective could be achieved by the integration of several portals of different research networks and the interfacing between these portals to maximize the usability and minimize the cost.
Stock Market Analysis
Analysing stock market data is a compute intensive task. Using heuristic approaches and secure distributed compute resources allows correlation predictions to be made in near real-time.
TIN Creation for a Finite Elemente Storm Flood Simulation in Hamburg
Flood simulation is a finite element calculation based on a flat water model. Basis for this simulation is a Triangulated Irregular Network (TIN) which is adapted to this special simulation calculation. The resolution of TIN can depend on the location and can be chosen in relation to flood relevant properties of the terrain like gradient, dike lines, and building density.
Starting point is a digital elevation model, consisting of measurement points with height information. To detect the important terrain properties automatically, a regular lattice with a mash size of a few meters is needed. For the huge Elbe area, the raster interpolation has to be done parallel is several sub areas. This sub areas are shown in the fig. below.
From this raster data parts, a unique TIN will be generated. This TIN will become the basis for the flood simulation calculation.
Simulation of Noise Immission in Dortmund
Detailed noise calculations with a spatial resolution of 10m can for large towns like Dortmund can only be done by splitting the town area into 1 × 1 km parts, which will be calculated in parallel. A map with such sub areas of Dortmund is shown below, the calculation time for the parts is shown in relation to the building density.