Iterative reconstruction of two-dimensional sinograms with TomoPy for magnetic particle imaging

Tomographic projection imaging technique is well recognized for its reliance on sinograms. Sinograms are the result of imaging scans in the field of medical imaging: CT, PET, etc. One of the more recent techniques that uses projection imaging is Magnetic Particle Imaging (MPI). In MPI, a subject with an injected nanoparticle tracer can be imaged using a magnetic field free line (FFL) at various angles like an x-ray beam in CT. Such imaging produces 2D sinograms that are generally reconstructed through inverse radon transformation. While this technique, coupled with filtered back-projection and deconvolution, approximates the subject's original characteristics effectively, it has some limitations that need to be improved in MPI: imaging artifacts and low spatial resolution. To address this objective in our MPI scanner, we opted to employ a hybrid iterative tomographic reconstruction in Python using TomoPy's libraries, standardly used for reconstructing 3D sinograms, to reconstruct a wide range of 2D data with a variety of iterative methods. Notably, this study introduces an effective method to manipulate 2D matrices and reconstruct their corresponding sinograms providing better image quality in MPI.