A Novel Transmission Reconstruction Algorithm for Radioactive Drum Characterization

Abstract

The accuracy of tomographic gamma scanning transmission reconstruction is a critical factor in reconstructing the activity of a radioactive drum. Traditional reconstruction algorithms produce severe grid artifacts and a high level of noise, thereby increasing the reconstruction error for both the density map and the activity. This paper proposes a novel algorithm for transmission reconstruction by combining maximum-likelihood expectation maximization and a convolutional neural network (CNN). Our experimental results indicate that the proposed reconstruction algorithm is capable of significantly reducing measurement errors, increasing spatial resolution while also eliminating grid artifacts, and being sufficiently robust when dealing with a noisy input image. The mean squared error of the output image decreased by 52.70% compared with the conventional reconstruction method, and the peak signal-to-noise ratio and structural similarity index improved by 21.89% and 17.33%, respectively. The spatial resolution was improved by 28 times, which demonstrates that CNN is a potentially useful new method for radioactive waste drum transmission image reconstruction.