Total: 1
This article presents a new robust model for image reconstruction in magnetic particle imaging (MPI) for single-axis drive field scans, which is based on the deconvolution of gridded harmonic data. Gridded harmonic data, used commonly in MPI, does not map to underlying iron density but rather to the iron density convolved with the harmonic point-spread functions. We refer to the gridded harmonic data as harmonic portraits, since they only represent a portrait-like representation of the iron density, and a deconvolution method is implemented to reconstruct the true underlying density. The advantage of this new method is primarily in the intermediate data analysis that comes in the harmonic portrait domain, where we are able to perform artifact correction, parameter selection, and general data assessment and calibrations efficiently. Furthermore, we show with several examples that our new method closely compares qualitatively with current state-of-the-art image reconstruction models in MPI. While the general concept of gridding harmonic data in MPI is not new, the complete modeling and characterization in order to use the data for image reconstruction has remained an ongoing area of research. We provide detailed analysis, theoretical insights, and many nuanced techniques that make our new methodology and algorithm accurate and robust.