Radiation imaging system based on the 32-FCNN and monolithic crystal detector

摘要: This paper proposes a radiation imaging system based on neural network ray positioning and a monolithic crystal detector. The monolithic crystal detector consists of a 5mm-thick monolithic LaBr3(Ce) crystal coupled with a SiPM array, demonstrating excellent system uniformity under 662 keV gamma-ray irradiation from Cs-137, with an energy resolution of 5.3%. By employing a 32-FCNN ray positioning model, the imaging system achieves an intrinsic spatial resolution of 2.67 mm, with an effective output area covering 81.3% of the detector surface, showing good positional linearity. The imaging capability of the system was evaluated through experiments. Results indicate that when an I-131 source is positioned 82 cm from the detector, with a focal length a of 10 cm (distance between detector and coding plate) and object distance b of 72 cm (distance between radiation source and coding plate), the reconstructed central point-source image using maximum likelihood estimation reconstruction algorithm with 20 iterations achieves a signal-to-noise ratio (SNR) of 14.2. Furthermore, the system exhibits a spatial resolution of 21 mm and an imaging range of ±11.2°.