Depth Of Interaction Estimation Research Articles

Depth of Interaction Estimation in a Preclinical PET Scanner Equipped with Monolithic Crystals Coupled to SiPMs Using a Deep Neural Network

The scintillation light distribution produced by photodetectors in positron emission tomography (PET) provides the depth of interaction (DOI) information required for high-resolution imaging. The goal of positioning techniques is to reverse the photodetector signal’s pattern map to the coordinates of the incident photon energy position. By considering the DOI information, monolithic crystals offer good spatial, energy, and timing resolution along with high sensitivity. In this work, a supervised deep neural network was used for the approximation of DOI and to assess through Monte Carlo (MC) simulations the performance on a small-animal PET scanner consisting of ten 50 × 50 × 10 mm3 continuous Lutetium-Yttrium Oxyorthosilicate doped with Cerium (LYSO: Ce) crystals and 12 × 12 silicon photomultiplier (SiPM) arrays. The scintillation position was predicted by a multilayer perceptron neural network with 256 units and 4 layers whose inputs were the number of fired pixels on the SiPM plane and the total deposited energy. A GEANT4 MC code was used to generate training and test datasets by altering the photons’ incident position, energy, and direction, as well as readout of the photodetector output. The calculated spatial resolutions in the X-Y plane and along the Z-axis were 0.96 and 1.02 mm, respectively. Our results demonstrated that using a multilayer perceptron (MLP)-based positioning algorithm in the detector modules, constituting the PET scanner, enhances the spatial resolution by approximately 18% while the absolute sensitivity remains constant. The proposed algorithm proved its ability to predict the DOI for depth under 7 mm with an error below 8.7%.

Evaluation of algorithms for photon depth of interaction estimation for the TRIMAGE PET component.

The TRIMAGE consortium aims to develop a multimodal PET/MR/EEG brain scanner dedicated to the early diagnosis of schizophrenia and other mental health disorders. The PET component features a full ring made of 18 detectors, each one consisting of twelve 8x8 Silicon PhotoMultipliers (SiPMs) tiles coupled to two segmented LYSO crystal matrices with staggered layers. In each module, the crystals belonging to the bottom layer are coupled one to one to the SiPMs, while each crystal of the top layer is coupled to four crystals of the bottom layer. This configuration allows to increase the crystal thickness while reducing the depth of interaction uncertainty, as photons interacting in different layers are expected to produce different light patterns on the SiPMs. The PET scanner will implement the pixel/layer identification on a front-end FPGA. This will allow increasing the effective bandwidth, setting at the same time restrictions on the complexity of the algorithms to be implemented. In this work two algorithms whose implementation is feasible directly on an FPGA are presented and evaluated. The first algorithm implements a method based on adaptive thresholding, while the other uses a linear Support Vector Machine (SVM) trained to distinguish the light pattern coming from two different layers. The validation of the algorithm performance is carried out by using simulated data generated with the GAMOS Monte Carlo. The obtained results show that the achieved accuracy in layer and pixel identification is above the 90% for both the proposed approaches.