用四次多项式插值消除医用X射线CT中的金属伪影

当今X射线CT技术已是医疗领域最重要的诊断手段之一，但实践中有时会碰到金属异物吸收X射线等问题，使得重构的图象上出现大量的条纹，因而干扰诊断，为了消除这种金属伪影，提出一种用四次多项式来以对缺失的投影值进行插值，并对重构的图象进行值裁剪的方法，该四次多项式是按若干准则确定的，即在投影缺口两端点处光滑过渡，并在缺口的内部尽可能保持线性，以使各方向上投影值之和为常数，而值裁剪是通过统计图象上灰度值的分布，然后剔除不合理的灰度值来实现的，实验表明，这种新方法极大程度地消除金属伪影，并能显著提高图象的对比度。

Reduction of Metal Artifact in X-ray CT by Quartic-Polynomial Interpolation

Nowadays X ray computerized tomography has been one of the most important diagnostic tools in medicine. However, sometimes the metallic implants absorb the X rays and therefore introduce streaks on the reconstructed images, which interfere diagnosis severely. In this paper, a simple interpolation method using quartic polynomials to fill in the missing projections is presented. The clipping on the pixel values of the reconstructed image is also proposed. The polynomial is determined by several criteria, namely smooth transition at the ends of the gap of missing projections, being as linear as possible inside the gap, and constant sum of interpolated projections in every direction. Value clipping eliminates irrational pixel values by evaluating the distribution of pixel values. Experiments show that by our method, the streaks caused by the metallic implants can be greatly suppressed and the contrast can be significantly improved.