The Erlangen National High Performance Computing Center accepted our application for additional compute resources for the MAGNET-Project

Approximately one year ago we started working on MAGNET4Cardiac7T - a research project together with the University Hospital in Würzburg, where we develop deep learning models for simulating electromagnetic fields, to enable patient-specific analysis of risk associated with using a 7T MRI scanner. Within the project we intend to use so called physics-informed neural networks (PINN), which integrate physical knowledge into neural networks. There are several challenges within the project that require large amounts of compute power. On one hand, this area of research is relatively new and requires a lot of experimentation with different neural network architectures. On the other hand the task we try to solve requires using a high resolution 3D image which is computationally very expensive.

For this reason we are very pleased with the decision to grant us the required computational resources to proceed with our experiments. Our application has been scientifically peer-reviewed and assessed as feasible in a technical evaluation. Based on these assessments our proposal has been fully approved.