Simultaneous estimation of the radioactivity distribution and electron density map from scattered coincidences in PET: A project overview

Abstract

Quantitatively accurate PET images require correction of measured data for scattered coincidences. Additionally, an anatomical image is required to provide accurate attenuation correction and to facilitate the interpretation of the activity distribution. By taking advantage of accurately measured photon energies and the kinematics of Compton scattering, a 2D surface described by two circular arcs (TCA), which define the possible scattering loci and encompasses the annihilation position, can be identified. In 3D the annihilation is confined to the volume encompassed by the surface obtained by rotating the 2D arc around its axis. Using this premise, we have developed novel iterative reconstruction algorithms which use the scattered coincidences to 1) improve the activity distribution and 2) obtain an electron density map. The results have demonstrated the feasibility and benefits of incorporating scattered coincidences into the image reconstruction process. Incorporating scattered coincidences directly into the radiotracer reconstruction algorithm eliminates the need for scatter correction, and could improve both image quality and system sensitivity. The electron density map reconstructed from scattered coincidences can be directly applied to attenuation correction of the activity distribution, which removes energy scaling and registration problems.