Optimized Data Acquisition for Image Reconstruction in Magnetic Particle Imaging (MPI) Based on Compressed Sensing

• Applicants: Professor Dr. Thorsten Buzug; Professor Dr.-Ing. Alfred Mertins
• Subject Area: Medical Physics, Biomedical Technology
• Term: from 2014 to 2019
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 250691157

Using Magnetic Particle Imaging (MPI) a local concentration of magnetic nanoparticles can be displayed quantitatively in real-time with high sensitivity and with very good spatial resolution. The basic idea is to utilize the non-linear magnetization characteristics of the particles which are used as tracers. For this purpose, the technique employs two magnetic fields, on the one hand a static selection field, on the other hand a dynamic alternating field. Once the nanoparticles are brought into the alternating field they produce a non-linear magnetization, which can be measured using a receive coil. Apart from the fundamental frequency of the alternating field the measured signal also contains harmonics, i.e. oscillations with multiples of the fundamental frequency, which is caused by the non-linearity. After separation of the harmonics from the applied basic signal the concentration of the nanoparticles can be determined. Spatial encoding is achieved using the static selection field. In first experimental studies MPI has already shown advantages over other imaging modalities. However, it has not yet reached its full potential with respect to spatial resolution, signal-to-noise ratio and acquisition times. It can be expected that recent advancements in signal processing and sampling theory, especially in the field of compressed sensing (CS), will contribute to the enhancement of image quality and speed. Sparse coding and compressed sensing have already improved other imaging modalities like e.g. Magnetic Resonance Imaging (MRI) considerably compared to the state-of-the-art of science. So far, the standard wavelet transformations have predominantly been applied as suitable transformations. Increasingly though transformations are sought that suit the signal characteristics of the respective modality. This strategy shall also be pursued in this project. Based on data of a simulation chain, which is to be developed, and realizations of different MPI scanner topologies by the Institute of Medical Engineering adapted transformations can be optimized for sparse coding, capitalizing on the expertise of the Institute for Signal Processing.

• DFG Programme: Research Grants