The ESPRIT funded European project CauchyPar constitutes a co-operative effort between five industrial and academic partners, in order to integrate the advanced Cauchy software, developed by the Rheinisch-Westfälische Technische Hochschule (RWTH), into a scalable parallel system. This high performance combination has to enable the localization and visualization of electrical discharges within the human brain. As a result, it is possible to derive mutual dependencies between the various discharges, and both normal and pathological functional regions in the brain. The more processors involved in the procedure, the less calculation time needed to locate the electrical activity.
In the nine month period from January'98 till September'98, the German partners Parsytec Computer GmbH, Philips Medizin Systeme, the RWTH Department of Neurology, and Biomagnetic Technologies have worked together with software experts from the University of Amsterdam - Computer Science, Physics and Anatomy Department at the Faculty of Mathematics to run the sequential Cauchy code on a parallel platform. The major goal was to prove that a higher performance of the code, which is based on a linear algorithm, can be extremely useful for real world applications in clinical environments.
During the first phase of the project, a parallel platform had to be selected with sufficient scalability and high performance to obtain the necessary processing speed. The choice fell on the Parsytec Cce-Systems because of the ease to transfer the code from a single-processor configuration to the parallel architecture by means of collision-free point-to-point communication at an elevated bandwidth. In this way, the amount of calculation time for the localization of electrical brain activity was brought back from a couple of days to a few hours or even minutes. In addition, the quality of the data and granular structure in the underlying model could substantially be optimized.The behaviour of the Cauchy software on the parallel environment has been tested extensively to receive a preliminary feedback of the HPC prototype's capacities. In the second phase of the project, the partners further improved the data flow and the overall distribution of processes, in order to maximize the visualization throughput. The combination of the CauchyPar software with the parallel processing system and the often readily available electro encephalographic (EEG) hardware at the neurological department in most hospitals provides an affordable solution for the measuring of electrical discharges within the brain.
The ChachyPar procedure outperforms alternative techniques, such as invasive operation methods or the use of expensive diagnostic hardware. Most important of all is that the patients far less have to be submitted too fairly dangerous surgical interventions. This initiative has equally been supported by the German division of the Transfer Technology Nodes (TTN). For further information, you can consult the home page of the CauchyPar project.