Magnetic Particle Imaging (MPI) is a new imaging technology that images the spatial distribution of iron oxide nanoparticles. Since the magnetic field strength that can be safely applied in MPI is limited, the field-of-view (FOV) must be divided into partial FOVs. Because the excitation magnetic field causes direct feedthrough on the receiver coil, the excitation frequency must be filtered out of the MPI signal. During this process, the nanoparticle signal at the same frequency is also lost, as a result of which each partial FOV experiences different levels of DC shift. In the standard MPI image reconstruction, these DC shifts are calculated from neighboring overlapping partial FOVs. Here, we propose a novel method that directly reconstructs the MPI image from the calculated DC shift values. Especially in the case of low bandwidth signal acquisitions, this method yields higher resolution images when compared to the standard method. The simulation results at various signal-to-noise ratios (SNR) show that the proposed method produces 6-8 dB increase in peak SNR and yields images that closely match the ideal image.