A Novel Open-Source Optimization Framework for the Design and Simulation of Radiofrequency Coils for Magnetic Resonance Imaging - Project Summary/Abstract A Novel Open-Source Optimization Framework for the Design and Simulation of Radiofrequency Coils for Magnetic Resonance Imaging Radiofrequency (RF) coils are essential in magnetic resonance imaging (MRI) because they affect the obtainable spatial and temporal resolution, the image homogeneity, and the encoding capability for parallel imaging, among other things. One limitation of the current approach to coil design is that the quality of a coil can only be compared against other available coils, giving no indication of whether there is room for further improvement beyond the best-performing design tested. Theoretical coil performance limits have been proposed as both absolute references to assess any coil design and target metrics for coil design optimization. The goal of this project is to develop and demonstrate a novel shape optimization approach for the design of RF coils based on ultimate performance benchmarks. This project will introduce a new method for rational coil design that relies on rapid volume-surface integral equation techniques to simulate the coils’ RF fields and advanced shape modeling algorithms to automatically explore geometric variations and find the optimal design. Starting from a target anatomical model, a coil substrate tailored to it, and a desired number of channels, our software toolkit will automatically optimize shape, position and geometric arrangement of the coil elements of an array to minimize the deviation of the simulated performance from the corresponding theoretical performance limit. This project will build upon computational electromagnetics methods recently introduced by our group for ultra-fast coil simulations and the calculation of ideal current patterns associated with optimal performance (ultimate intrinsic signal-to-noise ratio and ultimate intrinsic transmit efficiency) in realistic anatomical models. We will also rely on the extensive expertise of the project team with automatic mesh generation and shape optimization with adjoint- based gradient computation. We will validate the proposed coil design optimization framework in simulations and experiments. In particular, we will model, simulate, and then construct a flexible receive array for knee imaging at 3 tesla (T) and a 7 T transmit-receive head and neck coil. We will distribute all software as open source and fully documented, including tutorials and examples.
