体层合成的各向异性扩散滤波伪影消除算法

A Tomosynthesis Artifact Removal Algorithm Using Anisotropic Diffusion Filter

摘要:

体层合成(tomosynthesis)是一种有限角度图像重建技术，其成像角度范围比CT重建要小得多，这种成像结构产生了切片间的伪影。这种伪影在阅片时会影响对病灶的判断，从而影响诊断的正确率。已提出的投票策略可消除切片间伪影，但该算法依赖于分割算法，或依赖于足够多的投影数目进行估计，且得到的重建结果视觉过渡不自然。本文提出利用各向异性扩散滤波器来消除切片间的伪影，可有效克服投票策略存在的问题，实验验证了本文提出的算法的有效性。

Abstract: Tomosynthesis is a kind of limited-angle reconstruction with its acquisition angle range much less than CT, which leads to out-of-plane artifact. This kind of artifact can influence the decision of disease location during medical image reading, which deviates the accuracy of diagnosis. Voting strategy is proposed to eliminate out-of-plane artifact, but this algorithm depends on specific segmentation algorithm or estimation of sufficient number of project data. In addition, reconstruction result with voting strategy is abrupt to naked eyes. Anisotropic diffusion filter is used to remove out-of-plane artifact in this paper. The proposed algorithm in this paper can overcome the problems of voting strategy. The proposed algorithm is verified by experimental results in this paper.