Sensitivity Enhancement Using Triple-Coincidence Events in the AXPET Demonstrator

THE general goal of developments in detection systems for Positron Emission Tomography (PET) is the enhancement of image quality, generally without change to object activity. However, in an attempt to improve a particular detection parameter novel systems may measure new, information-bearing, data types that are often discarded or used only approximately in the process of image generation. One such example is the generation of increased levels of Inter-Crystal Scattering (ICS) that occurs in PET when detection resolution is enhanced [1]. ICS are potentially useful events that may be measured in a PET system yet are difficult to process and may introduce resolution degradation. In this investigation different methods of data treatment are explored withing the Maximum Likelihood Expectation Maximization (ML-EM) reconstruction algorithm to investigate the inclusion of ICS events directly into the system matrix. A novel approach [2] is compared to more standard methods of addressing triple-coincidence data. The AXPET prototype [3] was employed to acquire experimental data from a NEMA-like phantom [4] and standard figures of merit were calculated. The AXPET demonstrator uses a rotating-phantom geometry and suffers from long scan times and saturation effects at higher activities. While with great care the time-profile for each acquisition could be balanced, in this investigation we developed a method that could be used to approximate acquisition contributions from each phantom position to correct sensitivity estimation and hence enhance quantitative accuracy. Using experimental data from the AXPET and the more accurate sensitivity matrix, the method of ICS inclusion proposed in [2] is compared to standard approaches [5], [6] that only approximately model the transition probabilities of ICS in the system matrix. Finally, images of small animals, acquired using the AXPET demonstrator are used to illustrate ICS inclusion using biological systems.