2025/04/11 by Suskin, Philip, Förger, Fynn, Jürß, Paul +3
#Computer Science #Electronic Engineering #Information and General Works::004: Computer Sciences #Natural Sciences and Mathematics::530: Physics #Technology::621: Applied Physics::621.3: Electrical Engineering
paper · doi:10.15480/882.15068
The scaling of electrical power constitutes a significant challenge when adapting Magnetic Particle Imaging (MPI) to a human scale. The use of coils incorporating soft-iron cores serves to reduce power usage, but also introduces spatial imperfections and non-linearities in the current-to-field relationship. This study proposes methodologies for the control of the magnetic field output of a system comprising 18 coils, subject to the influence of saturated iron. In particular, we integrate current sequence optimization with neural network-based predictions for field and gradient values, thereby enabling the precise and power-optimal generation of magnetic fields. The proposed framework for controlling non-linear magnetic field generators represents a significant advancement in MPI technology, paving the way for the development of human-scale, power-efficient medical imaging solutions.