A Fast Crystal Identification Algorithm Applied to the LabPET™ Phoswich Detectors

Detectors based on LYSO and LGSO scintillators in a phoswich arrangement coupled to an avalanche photodiode are used in the LabPETtrade, an all-digital positron emission tomography (PET) scanner for small animal imaging developed in Sherbrooke. A Wiener filter based crystal identification (CI) algorithm achieving excellent discrimination accuracy was recently proposed for crystal identification of LYSO-LGSO phoswich detectors . This algorithm was based on estimating parameters describing the scintillation decay time constant and the light yield of events sampled at 45 MSPS. The CI process was performed by applying a threshold on the scintillation decay parameter of events. The light yield was not considered in the CI process even if it should be. We propose a 2-fold faster CI approach which takes both the scintillation decay and light yield coefficients of each crystal into consideration. The new algorithm uses the previous Wiener filter based algorithm as a calibration process in order to evaluate the model of each individual crystal. The DAQ chain model as a priori knowledge is then incorporated into the model of each crystal and the output signal is estimated. The CI is performed by evaluating a single parameter representing the percentage contribution of each crystal characteristics in the event signal. The CI algorithm demonstrated a discrimination rate accuracy for LYSO-LGSO LabPET detectors and for LSO-GSO crystals in phoswich arrangement for 511 keV events. Although a calibration is required, the real-time implementation of the new CI algorithm needs 2 times less direct operations. An FPGA clocked at 400 MHz can process up to 25 M events/sec with such an algorithm.