Compensating model imperfections during image reconstruction via Resesop

Abstract

To avoid the time consuming process of measuring the system function of magnetic particle imaging, model-based system matrix simulation is an alternative. However, this is a complex procedure with model imperfections, which influence the accuracy of the resulting system matrix as well as the quality of the image reconstruction. Standard reconstruction algorithms like regularized Kaczmarz are not able to take those inexactnesses into account and produce poor quality images. The Resesop-Kaczmarz algorithm is a novel image reconstruction method, which allows model imperfections or dynamics. We examine and discuss the compensating characteristics of Resesop-Kaczmarz regarding model inexactnesses.