高速中性子線を用いた高性能ＣＴ再構築アルゴリズムの開発

为了实现光学、X射线和中子等不同成像方式的超分辨率CT成像，提出了一种线性+旋转扫描的方法，在每个视角采集更多的投影数据进行图像重建。本文首先以快中子CT为研究对象，采用像素尺寸为2cm×2cm×5cm（T）的像素型快中子成像探测器。编制了实现面积模型下系统矩阵计算的程序。投影数据和系统矩阵的重建结果初步表明，线性+旋转扫描方法比旋转扫描方法能更好地改善CT图像质量，为了进一步验证线性+旋转扫描方法在改善CT图像质量方面的有效性，在RIKEN加速器驱动紧凑型中子源（RANS）上进行了快中子CT实验。对1.5cm厚的亚克力棒的成功识别表明，线扫描方法在一张投影图像上效果良好，针对目前像素型快中子成像探测器CT图像质量差，难以评价其空间分辨率的问题，采用热中子CT对CT图像的空间分辨率进行评价。利用模拟结果实现了通过目标锐边的半高宽（FWHM）评价空间分辨率，并在第三届能源与环境先进测量、分析与控制国际研讨会（AMACEE 2020-WEB）上就像素型快中子成像探测器的性能评价做了口头报告（特邀报告）。

To realize the super resolution computer tomography (CT) with different modalities, such as optics, X-rays and neutrons, a linear + rotation scan method is proposed to collect more projection data in each view angle for image reconstruction. We firstly set fast neutron CT as the research object, which utilizing a pixel-type fast neutron imaging detector with 2cm×2cm×5cm (T) pixel size. We have made the code to realize calculation of system matrix with area model. Reconstruction results with projection data and system matrix preliminary indicates linear + rotation scan method can improve the quality of CT image compared with rotation scan.To further validate the effectiveness of the proposed linear + rotation scan method in improving the quality of CT image, fast neutron CT experiment has been done at RIKEN Accelerator-driven compact Neutron Source (RANS). Successful identification of 1.5cm thick acrylic bar indicates linear scan method works well in one projection image.As the spatial resolution of CT image with current pixel-type fast neutron imaging detector is hard to be evaluated due to its weak image quality, thermal neutron CT is used to evaluate the spatial resolution of CT image. Evaluation of spatial resolution through Full Width Half Maximum (FWHM) for sharp edge of object with simulation results has been realized.We have made an oral presentation (invited talk) about evaluation of performance of pixel-type fast neutron imaging detector on 3rd International Symposium on Advanced Measurement, Analysis and Control for Energy and Environment (AMACEE2020-WEB).