Improving the accuracy of PEPT algorithms through dynamic parameter optimisation

Positron emission particle tracking (PEPT) is used to study a wide range of scientific, industrial, and biomedical systems, typically those inaccessible through conventional optical particle tracking techniques. However, in dense or thick-walled systems a fraction of the coincident gamma-rays emitted from a PEPT tracer, called Lines-of-Response (LoRs), are attenuated via Compton scattering. Additionally, at high source activity, random LoRs may be formed by two unrelated events. The incorporation of scattered or random LoRs decreases PEPT spatial accuracy and can distort the trajectory. In this work, we use validation experiments and simulations to investigate the spatial accuracy of the Birmingham Method (BM) PEPT algorithm when two key free parameters are changed: the total number of LoRs in the sample and the fraction of LoRs in the sample used to locate the tracer. Our results show that the default algorithm parameters are not suitable for all cases, however, Monte Carlo simulations of PEPT experiments can be used to estimate the optimal parameter values. Ultimately a variant of the BM, called Dynamic-BM, is demonstrated in a virtual PEPT experiment. Dynamic-BM uses the optimal parameters on a sample-by-sample basis improving PEPT accuracy in this case by 4.03% over the best constant parameters and 76.5% over the default parameters. These improvements make PEPT a more accurate and thus more useful tool. •Two key free parameters a PEPT algorithm are optimised for spatial accuracy.•Optimal parameter values are discovered through experiments and simulations of PEPT.•The Birmingham Method is extended to dynamically change parameter values.•Using optimal values, spatial errors decrease by 76.5% compared to default values.