Computationally efficient image reconstruction via optimization for X-space MPI

Magnetic Particle Imaging (MPI) is a promising new modality that images only a magnetic tracer, commonly super-paramagnetic iron oxide (SPIO) nanoparticles. During data acquisition information about the DC component of the native image is lost, and must be recovered using some a priori knowledge such as image continuity and positivity. One method of recovery uses convex optimization techniques to determine the DC offsets that match the a priori knowledge. This method produces excellent images but significant time and memory resources to perform. This paper discusses an efficient numerical software package, PyOp, for performing optimizations with greatly reduced space and time taken to perform a reconstruction. As an example, this package is applied specifically to x-space MPI DC reconstruction, allowing the DC reconstruction to both run faster on even modest computing hardware.