An improved nearest neighbor method for the estimation of the gamma photon entry point in monolithic scintillator detectors for PET

Abstract

Several improvements of the k-nearest neighbor (k-NN) method for the determination of the entry point (x, y) of a gamma photon in a monolithic scintillator PET detector have been investigated with the aim to obtain better spatial resolution and/or to enable faster detector calibration by reducing the amount of required reference data and by allowing for calibrating with a line source. These methods were tested on a dataset measured with a SiPM-array-based monolithic LYSO detector. It appears that ∼10% to ∼25% better spatial resolution can be obtained compared to the standard approach. Moreover, some of the improved methods using two orders of magnitude less reference data, yield essentially the same spatial resolution as the standard method, which reduces the time needed for calibration as well as entry point computation. Finally, line source calibration is shown to be possible with some of the methods, yielding better results than the standard method and allowing much faster and easier collection of the reference data.