CT Image Reconstruction Algorithm to Reduce Metal Artifact

Abstract

Micro-computed tomography (µCT) is quite useful for nondestructive evaluation of devices with complicated internal structures. However, large difference in X-ray absorption coefficients of materials produces streak and star pattern artifacts in the CT images. The artifact, which is called “metal artifact”, makes it difficult to inspect the image of the device. This study aims to develop a new reconstruction method to reduce the artifacts from the images. In this paper, we firstly point out the cause of the artifacts using CT data of a sample. Next, we propose a CT reconstruction algorithm to reduce the artifacts. The basic idea of the algorithm is to correct the projection data. Shapes of high density parts such as metals are extracted by a threshold method and projection data are replaced to consistent values. We apply the method to some devices and confirm availability for the nondestructive inspections.