Validation of 3D model-based maximum-likelihood estimation of normalisation factors for partial ring positron emission tomography

Abstract

The next generation of organ specific Positron Emission Tomography (PET) scanners, e.g. for breast imaging, will use partial ring geometries. We propose a component-based Maximum-Likelihood (ML) estimation of normalisation factors for 3D PET data reconstruction applicable to partial ring geometries. This method is based on the Software for Tomographic Image Reconstruction (STIR) for full ring PET and is validated for a stationary partial ring scanner. The model includes the estimation for crystal efficiencies and geometric factors. The algorithm is validated using Maximum Likelihood Estimation Method (MLEM) based 3D reconstruction in STIR using Geant4 Application for Tomographic Emission (GATE) simulation data for full and partial ring scanners and experimental data from a demonstrator with partial ring geometry. The uniformity of the reconstructed images of simulated cylindrical and NEMA-IQ phantoms in both scanner geometries and the image of a line source in the partial ring demonstrator is assessed. The results have shown that uniform images in both axial and transaxial directions are obtained after applying the estimated normalisation factors. The accuracy of the algorithm is validated by comparing the normalisation factors between the full and partial ring systems in simulation. We have shown that the estimated normalisation factors are almost identical, even though the separate components are not. This proves that the ML estimation of the 3D normalisation factors is valid and can be applied to the partial ring scanner.