Project B15 - Temporal regularization for MPI Reconstruction
Principal Investigator: Christina Brandt
Background and Motivation
Magnetic Particle Imaging, short MPI, is a novel imaging technique which uses the signal generated by the nonlinear response of magnetic nanoparticles to an inhomogeneous magnetic field. The concentration of a magnetic tracer in the patient‘s body is determined to examine physiological processes. MPI reconstruction is an ill-posed inverse problem. In general it is not possible to find a unique solution but there are methods to approximate a solution. Regularization raises conditions for the solution and stabilizes the method. Due to the speed of signal acquisition and the aim to visualize physiological processes MPI is well suited for the acquisition of time series. Furthermore the field of view of MPI is limited so that larger volumes are a composition of smaller subvolumes (patches) which are scanned sequentially. Therfore there is a time dependence even for static images.
Aims and Objectives
The objective of this project is to find regularization methods for the reconstruction of time series. We will try to use the redundant information or similarities of subsequent measurements for regularization. This might have an impact on the quality of the method regarding stability, speed or image quality. It will be examined how the information of the time dependent data be used for regularization. As a first approach methods applied to the problems of dynamic CT or PET will be transfered to the problem of MPI reconstruction. For evaluation a simulation framework which can work with data from real and simulated MPI measurements will be developed. It will allow to test the developed methods under different conditions like static and multipatch images as well as time series containing different kinds of motion.
PhD Student: Christiane Schmidt