Event Pulse Classification with Multi-Voltage Threshold samples for scatter rejection in PET

Abstract

We proposed a novel Event Pulses Classification (EPC) method to reject scatter directly from Multi-Voltage Threshold (MVT) samples data. This study is motivated by the emerging demand for the MVT-digitalized scintillation pulses processing in the PET implementation. MVT is a sparsequantization-level flash ADC which can digitize fast signals (such as scintillation pulses without shaper) with excellent budget. However, it is limited by the absence of integral expression and baseline, leading to difficulties in energy-related processing, such as scatter rejection and event location. Based on Maximum a Posterior (MAP) criterion, EPC introduces a new classifier technology to overcome the above difficulties. To evaluate the performance of our EPC method in the process of exploring useful information of MVT sampling, we set up our detection platform and collect a dataset of event pulses. Events in this digital library are obtained at a high sampling rate of 50 GSps (Giga-samples per second) so that their waveforms are recorded with high accuracy. Our results show that our EPC method indeed yield a very low Mean Error Rate (MER) with MVT samples. Meanwhile, these results of EPC have also demonstrated the promising potentials of MVT techniques in the energy-related classification of pulses.