基于重建图像全角度前投影的硬化校正方法

Beam-Hardening Calibration Based on Full Forward Projection of Reconstructed CT Image

现有的硬化效应校正方法依赖于射线频谱等先验条件,校正过程较为复杂,因此提出一种基于重建图像全角度前投影的硬化校正方法.对重建后的图像进行区域分割,提取出杯状伪影主要影响的组织区域;然后进行基于像素点的全角度前投影,获得校正基算子;再将校正基算子及其高次方乘积进行线性组合获得校正算子,并将此校正算子应用于原图像从而达到校正的目的.水模、头模和西瓜的校正结果显示,本方法对单物质及近人体的物质重建图像的硬化效应均有较好的校正效果.

The existing calibration methods commonly rely on the X-ray spectrum or other priori knowledges. So, a new beam-hardening calibration method is proposed on the basis of the full forward projection of a reconstructed image. First, a reconstructed image is segmented according to its gray values to obtain the tissues involving cupping artifacts. Then, a basic corrector is achieved through the full forward projection based on the pixels. By linearly combining the transformations of the basic corrector, a final corrector is obtained to make up for the cupping artifacts. The results of correcting the beam-hardening in single material and soft tissues like water phantom, head phantom and watermelon demonstrate that the proposed approach is effective.