A Feaibility Study of 3D Cosmic-Ray Muon Tomography with a Portable Muography Detector

Abstract

The Maximum Likelihood-Expectation Maximization (ML-EM) method was applied to 3D image reconstruction of cosmic-ray muon tomography. The feasibility was examined by using Monte Carlo simulation for a simple configuration where two lead blocks were placed at a different height from a muography detector. The 2D projection of the average thickness of the blocks as a function of the muon direction was simulated for multiple detection positions. The 3D image of the density profile was reconstructed by applying the ML-EM method to the simulated projections. It was found that the image reproduces reasonably well the position of the two blocks. The effect of the limited number of detection positions and the number of iteration in the ML-EM method on the image reconstruction was investigated in detail.

3D cosmic-ray muon tomography using portable muography detector

“…In the present work, we propose an approach based on the ML-EM algorithm [12] for cosmic-ray muon tomography. A simulation was developed to test the feasibility of the algorithm [13] and designed the experiment's configuration using our developed portable muography detector [10]. Then, the 3D muography image from the experiment setup was also performed to confirm the feasibility.…”

“…Since the ideal detector for 3D tomography has a solid angle of 4𝜋 sr from the target, as Radon's theorem [13] proved, the target shape can be determined uniquely under this condition. However, as mentioned in the Introduction section, the effects of low cosmic-ray muon flux and its angular distribution on the measurement duration and the number of detection positions are significant constraints in muography.…”

“…However, the iteration stopping criterion is required for the acceptable quality. In this study, a simulation code was developed to estimate the number of iterations (see [13] for detail).…”

“…The detection positions, measurement times, and the ML-EM convergence conditions was determined by the simulation (See ref. [13] for the detail).…”

“…ratio maps were calculated for all eleven positions using a Monte Carlo method [13]. The number of simulated events was 10 7 muons for each detector position, with a statistical uncertainty of less than 1% for all of the pixel of absorption ratio maps.…”