Comparison of Reconstruction Methods for Elongated Multi-Patch Sequences in Magnetic Particle Imaging

Abstract

Magnetic particle imaging systems utilize one or more dynamic drive fields to excite the magnetization of the nanoparticles, along with a static selection field that spatially encodes the resulting magnetization signal. This selection field effectively suppresses signal generation outside a smaller volume surrounding the field-free region (FFR), while the drive fields swiftly move this region. However, the practical size of the possible field of view attainable with these single-patch sequences is limited. To upscale an MPI system, one can increase the effective field of view through two methods: moving the object or utilizing additional low-frequency focus fields. These extra fields can continuously or stepwise relocate the position of the FFR. Especially elongated sequences that involve slow and continuous shifts of focus fields along with simultaneous movements of drive fields in the FFR are of particular interest due to their flexibility and operability at full duty cycle. For reconstruction one either considers the entire sequence as a single patch or as a collection of multiple patches, each of which does impact image artifacts, reconstruction time, and calibration complexity. Dependent of the interpretation of the raw data particular challenges include the processing of the measured data, which is no longer periodic, and the length of the data sets that is now determined by the duration of the focus field sequence rather than by the comparably short drive field excitation cycles. In this work, various methods for reconstructing elongated multi-patch sequences are examined and contrasted based on their impact on image artifacts, runtime, and calibration complexity.