Sub-Frame Chebyshev Reconstruction for Magnetic Particle Imaging

Abstract

Magnetic Particle Imaging typically performs image reconstruction only after the full measurement has been acquired. This work investigates whether partial reconstructions during an ongoing measurement can be advantageous when only sets of sub-frames of the acquired signal are used. Static particle concentrations and their corresponding voltage signals were simulated, segmented using Hann windows, and reconstructed via direct Chebyshev reconstruction followed by kernel-based deconvolution, and cumulative reconstructions were obtained by summing the sub-frame results. The results show that reconstruction quality depends strongly on the regularization parameter and the information content of the selected sub-frames. For large regularization values, the full signal yielded the lowest error, whereas for sufficiently small regularization parameters, certain window configurations outperformed the full-signal reconstruction. These findings suggest that selectively excluding low-information sub-frames may improve reconstruction stability and reduce computational effort.