Lifton J J
J Xray Sci Technol. 2017;25(4):629-640. doi: 10.3233/XST-16197.
Beam hardening causes cupping and streaking artifacts in computed tomographic images, the presence of such artifacts can impair both qualitative and quantitative analysis of the reconstructed data. When the scanned object is composed of a single material, it is possible to correct beam hardening artifacts using the linearization method. However, for multi-material objects, an iterative segmentation-based correction algorithm is needed, which is not only computationally expensive, but may also fail if the initial segmentation result is poor. In this study, a new multi-material linearization beam hardening correction method was proposed and evaluated. The new method is fast and implemented in the same manner as a mono-material linearization. The correction takes approximately 0.02 seconds per projection. Although facing a potential disadvantage of requiring attenuation measurements of one of the object's constituent materials, applying the new method has demonstrated its capability for a multi-material workpiece with substantial reduction in both cupping and streaking artifacts. For example, the study showed that the absolute cupping artefacts in steel, titanium and aluminum spheres were reduced from 22%, 20% and 20% to 5%, 1% and 0%, respectively.
束硬化会在计算机断层扫描图像中导致杯状伪影和条纹伪影,这些伪影的存在会损害重建数据的定性和定量分析。当扫描对象由单一材料组成时,可以使用线性化方法校正束硬化伪影。然而,对于多材料对象,需要一种基于迭代分割的校正算法,这种算法不仅计算成本高,而且如果初始分割结果不佳可能还会失败。在本研究中,提出并评估了一种新的多材料线性化束硬化校正方法。新方法速度快,并且以与单材料线性化相同的方式实现。校正每个投影大约需要0.02秒。尽管面临需要测量对象的一种组成材料的衰减这一潜在缺点,但应用新方法已证明其对多材料工件的校正能力,杯状和条纹伪影都大幅减少。例如,研究表明,钢、钛和铝球中的绝对杯状伪影分别从22%、20%和20%降至5%、1%和0%。