CT Image Reconstruction Algorithm to Reduce Metal Artifact

Abstract

X-ray CT reconstructs cross-sectional image of an object using projection data collected from many different directions. This technology has spread rapidly in the field of diagnostic medicine as a low-invasive inspection. X-ray CT has been brought great benefit to us, however, it is still facing some technical issues. In particular, metal artifact is a significant problem in CT imaging. Metal artifact is a strong streak noise caused by a discrepancy of projection data, and makes it difficult to diagnose patients with metal implants in his/her body. Even though various studies to reduce metal artifacts were conducted, the problem is not resolved yet.

In this study, we propose a reconstruction algorithm to reduce metal artifacts. The algorithm is based on the nature of sinogram which is also known as the Radon transform in rectangular coordinates. The sinogram is a data that is arranged projection data from many different directions. By this definition, sinograms consists of piles of sine curves and our algorithm utilizes this feature to resolve the discrepancy of the sinogram.

We developed the CT simulation software in order to verify effects of complex situations assuming actual medical diagnoses. As a result, metal artifacts were effectively reduced either.