Modelling the GE discovery 690 PET/CT scanner

In this paper, the PET subsystem of the GE Discovery 690 PET/CT scanner is modeled in order to construct the system matrix which is a prerequisite for many iterative reconstruction methods. The system matrix, H is a representation of the projection of the object onto the scanner's detector. The projection of the image (I) onto detections (S) is thus represented as a linear equation, S = H I. Image reconstruction is basically solving an inverse problem and iterative image reconstruction is a powerful solution. As a demonstration of its effectiveness we use the basic Algebraic Reconstruction Technique (ART) with the built model of the scanner. As a comparison, we show how using analytic techniques such as filtered backprojection (FBP) is complicated for scanners of complex geometry.