The dual head panel PET image reconstruction based on simulated system response matrix

Abstract

Measuring system response matrix (SRM) by Monte Carlo method is time and resource consuming, even with high performance computer nodes. In this paper, we exploited symmetry properties available to drastically reduce the complexity in computing SRM of panel PET, which has two parallel panel detectors. By extending the original system and simulating 4 voxels every slice parallel to the PET detector, SRM of other voxels on the same image slice can be obtained by symmetry and translation. Without loss of generality, the voxel size in Y and Z direction is one-quarter of the crystal pitch size respectively, and they may not be the same. The results show that the noise level and convergence of reconstruction images by Monte Carlo method and symmetry are better than that by calculating SRM on-the-fly like solid angle.