Main Article Content
Copyright (c) 2021 Jorge Chacon-Caldera, Heinrich Lehr, Kulthisa Sajjamark, Jochen Franke
This work is licensed under a Creative Commons Attribution 4.0 International License.
Enhancements in spatial resolution can open new avenues for novel applications, but acquiring data at higher resolutions generally comes with penalties in measurement times, signal-to-noise ratios and safety concerns. Therefore, maximizing the spatial resolution of the available data during image reconstruction is paramount. Magnetic Particle Imaging (MPI) has already reached sub-millimeter spatial resolutions. With standard tracers, this has been achieved using a reconstruction method that compensates for the point spread function of the system and the superparamagnetic iron oxide particles (SPIOs). This method is known as the system matrix approach and uses a calibration measurement, relating the concentration of SPIOs and the true particle response to the measured signal. Using a calibration measurement for reconstruction requires a comprehensive assessment of the quality of the system matrix in addition to the measured image data. Analyzing the system matrix by reconstructing selected measurements contained in itself and visualizing them in image space (henceforward called eigen-reconstructions) can provide clear information regarding image quality and artifacts. This is equivalent to using ideal measurement data. Thus, it is possible to identify sources of image artifacts arising solely from the reconstruction, which can be then compensated. In a preliminary report, we presented the principle of eigen-reconstructions to identify and reduce reconstruction-induced artifacts. In this work, we focus on the application of the method to enhance spatial resolution in MPI reconstructions. The principles of an iterative algorithm based on eigen-reconstructions are also further detailed. It is shown that the algorithm compensates the blur arising during image reconstruction effectively. For validation, we present tests of our method using 2D and 3D datasets including an homogeneous and a resolution phantom to demonstrate potential opportunities and limitations.
Int. J. Mag. Part. Imag. 7(2), 2021, Article ID: 2112003, DOI: 10.18416/IJMPI.2021.2112003