A simulation study of a fan-beam time-of-flight fast-neutron tomography system. (July 2019)

• Record Type: Journal Article
• Authors: Sun, Shifeng; Ouyang, Xiaoping
• Is Part Of: Applied radiation and isotopes. Volume 149(2019:Jul.)
• Page Start: 52
• Page End: 59
• Publication Date: 2019-07
• Subjects: Fast neutron tomography -- Time of flight -- GEANT4 -- Image reconstruction; Radiology -- Periodicals; Radiation -- Industrial applications -- Periodicals; Nuclear chemistry -- Periodicals; Internet resource; Periodical; 660.298
• DOI: 10.1016/j.apradiso.2019.04.017
• Languages: English
• ISSNs: 0969-8043

Abstract

Abstract: A fast neutron imaging system, consisting of a deuterium-tritium neutron generator, an associated alpha-particle detector and an arc-shaped array of individual neutron detectors is simulated for fan-beam fast neutron tomography purpose. In this system, the associated particle and time of flight techniques are used to separate transmission neutrons from scattered neutrons by electronically collimating and time tagging the source neutrons. To evaluate factors affecting the system's spatial resolution and image quality, a series of simulations were performed with a GEANT4 based Monte Carlo program. The parameters include neutron detector resolution, system coincidence resolving time and reconstruction algorithm. The simulation results indicate that a reconstructed image resolution of about 2.1 mm can be obtained with an array of two hundred 4 mm resolution neutron detectors. With the aid of electronic collimation and time tagged based event filtering, the reconstructed image quality was significantly improved. Among iterative reconstruction algorithms tested, the Simultaneous Algebraic Reconstruction Technique-Total Variation algorithm showed better overall performance than others. Highlights: GEANT4 based Monte Carlo simulation of a fan-beam time-of-flight fast neutron tomography system. Images of test phantom made using an array of two hundred neutron detectors indicate that a 2.1 mm resolution is achievable. Reconstructed images using total variation based image reconstruction algorithms had the best quality. Results show that electronic collimation and time tagged event filtering could significantly improve the image quality.