Evaluation of PET positioning algorithms

Abstract

Images reconstructed in positron emission tomography(PET) depends on determining the position of a gamma on a scintillator array from the body of interest. For this project a highly pixelated scintillator array is coupled to a significantly less pixelated photodetector array (SiPM) via a light guide that spreads that spreads a shower of scintillation photons over the latter. Most commonly, Centre-of-Gravity algorithms are used to estimate the position of the initial photon on the scintillator based off the distributed readout from the SiPM. Under this study the different variations of the centre of gravity algorithm discussed are the standard centre of gravity algorithm, the intensity-weighted centre of gravity algorithm, and the iterative-weighted centre of gravity algorithm. Success is determined by comparison of flood histograms and the ability to render scintillator crystal centres based off of information from a silicon photomultiplier array. Further evaluations are to be made by comparison of different powers of the weight in the intensity-weighted centre of gravity, and of multiples of the Gaussian width in the iterative centre of gravity. Based on the data evaluated in this project it is difficult to say whether the intensity weighted and iterative position weighted centre of gravity models are able to present a significant improvement to the standard centre of gravity based on evaluation of the flood histograms alone. For the intensity weighted centre of gravity, a weight power of 1.2 is found to be optimal. No significant difference for the gaussian width in the iterative weighted centre of gravity is found.