Neutron displaced CT (computed tomography) scanning is an effective tomography detection method for large-sized samples, but the truncated projection data leads to significant calibration errors of the center of rotation (COR) of the turntable in the CT system, seriously affecting imaging quality. We consider the COR calibration error during the design of the neutron displaced CT scanning imaging method. A COR calibration algorithm of the turntable under the displaced CT scanning is designed. Then, the symmetric complementary data (SCD) reconstruction algorithm and the projection data preprocessing (PDP) reconstruction algorithm are established. The sensitivities of the reconstruction accuracy of the two reconstruction algorithms to the COR calibration error are discussed. We hope that the proposedCOR calibration algorithm and the reconstruction algorithm under the neutron displaced CT scanning mode can lay a theoretical foundation for solving the neutron CT imaging problem of large-sized samples.

A preciseCOR calibration method under the neutron displaced CT scanning mode is established. The calibration algorithm based on the symmetry principle of projection data is designed. Each possible COR position is enumerated, and the variances between the sum of the projection data on the left and the right sides of the COR are calculated. Finally, the COR result is determined by finding the location where the variance has the minimum value. Under the displaced CT scanning mode, the truncation and redundant characteristics of projection data will result in bright circular artifacts in the reconstructed images. We design two reconstruction algorithms to eliminate the bright circular artifacts, namely SCD reconstruction algorithm and PDP reconstruction algorithm. SCD reconstruction algorithm supply the missing projection data under displaced scanning mode by using the principle of symmetric complementary data and then use the filtered back projection (FBP) algorithm to obtain the accurate reconstruction result. PDP reconstruction algorithm utilize the WANG weighting function to process sinogram data. In order to eliminate the bright circular artifacts in the reconstructed image, the redundant projection values are weighted to ensure that the projection data from all directions contribute the same data amount to the reconstruction results. A simulation method of neutron projection noise including Gaussian noise and γ white spot noise is proposed, and a 3D simulation phantom is designed to verify the performance advantages of the proposed COR calibration algorithm and PDP reconstruction algorithm under different COR displaced sizes and projection noise intensities. A neutron displaced CT scanning imaging experiment is conducted based on the reactor neutron source to verify the practicality and stability of the proposed COR calibration algorithm and the reconstruction algorithm.

By using the designed 3D simulation phantom, it can be verified that the proposed COR calibration algorithm has a calibration error of 0.1. After adding Gaussian noise and γ white spot noise to the projection data of the 3D simulation phantom, the noise in the projection image is similar to the actual neutron data. As the noise intensity increases, the COR calculation error of the OAC (opposite angle calibration) algorithm significantly increases, but the COR calibration error of the proposed method does not increase (Table 2). Therefore, it can be proven that the proposed COR calibration algorithm has higher accuracy and stability. When the COR displaced size changes, the calibration error of the COR does not significantly increase. When the COR calibration error reaches two pixels, the reconstruction results of SCD reconstruction algorithm show certain image artifacts, resulting in distortion of the detailed structure in the reconstructed image. Additionally, due to the influence of stitching and misalignment, the reconstructed image also shows certain stripe artifacts (Fig. 8). The reconstruction results obtained by PDP reconstruction algorithm have stronger image detail resolution and higher reconstructed image quality (Fig. 9). When the projection has Gaussian noise and γ white spot noise, the reconstruction results obtained by PDP reconstruction algorithm are also better than those obtained by SCD reconstruction algorithm (Fig. 10). Moreover, PDP reconstruction algorithm can also achieve good reconstruction results when the COR displaced size changes (Fig. 11). Based on the reactor neutron source of China Academy of Engineering Physics, a neutron displaced CT scanning experiment is carried out, and clear internal and external structural details of the sample are obtained. The imaging field of the neutron CT system is expanded by 31.4%.

We design a neutron displaced CT scanning imaging method for large-sized samples and a COR calibration algorithm under the neutron displaced CT scanning based on the symmetry principle of projection data. The proposed COR calibration algorithm has the advantages of high measurement accuracy and strong anti-noise ability. Two neutron displaced CT scanning reconstruction algorithms are developed. SCD reconstruction algorithm is more sensitive to COR calibration errors. A smaller error can lead to stitching and misalignment issues in the supplemented projection data, thereby affecting the quality of image reconstruction. PDP reconstruction algorithm has a strong tolerance for the COR calibration error and can obtain higher reconstructed image quality. The 3D simulation phantom verifies the performance advantages of the proposed calibration algorithm and PDP reconstruction algorithm under different COR displaced sizes and projection noise intensities. In addition, the neutron displaced CT scanning experiment prove that the proposed COR calibration algorithm and PDP reconstruction algorithm have significant engineering practical values, laying a theoretical foundation for solving the problem of neutron CT imaging of large-sized samples.