PEPT-EC, a location algorithm based on anomaly detection for positron emission particle tracking (PEPT)

Positron Emission Particle Tracking (PEPT) has provided insights into the dynamical behaviour of mineral processing equipment and many other applications where optical tracking is not possible. Highly precise and frequent measurements are needed in order to measure the location of the tracer particle with time. This work presents a new location algorithm called "PEPT-EC" which is based on anomaly detection methods and tests its performance for tracking a tracer particle in comparison to two existing algorithms, Birmingham and PEPT-ML. The raw data for the measurements were generated from a Monte Carlo based simulation of a tracer particle moving on a circular path of radius 2.0 mm. PEPT-EC produced location data with lower or equivalent uncertainty in the path to the two other algorithms, with a considerable improvement at low activity. It achieved this with lower processing time than PEPT-ML, which is likely related to the reduced complexity of removing outliers compared to finding the centre of a cluster with more data points. PEPT-EC facilitates precise tracking of smaller tracer particles than is routine for PEPT with the potential to measure sub-millimeter flow paths in fine particle flotation in mineral processing and cellular transport in medicine.