CAJ | 학술논문

针对癌变CT图像中各器官变形严重，传统分割算法无法有效、完整地分割整个腹腔的问题，提出基于三维网格的分割算法。首先借助肋骨和脊椎等骨架获得初始腹腔骨架；然后构造一个球形初始三角网格，并建立与腹腔骨架间的关联；再通过3个目标函数变形网格；最后在网格附近的边界点优化网格，获得腹腔分割结果。实验结果表明，该算法能够有效地对病变严重的CT图像进行腹腔分割，对噪声和器官变形有良好的鲁棒性。
침대암변CT도상중각기관변형엄중，전통분할산법무법유효、완정지분할정개복강적문제，제출기우삼유망격적분할산법。수선차조륵골화척추등골가획득초시복강골가；연후구조일개구형초시삼각망격，병건립여복강골가간적관련；재통과3개목표함수변형망격；최후재망격부근적변계점우화망격，획득복강분할결과。실험결과표명，해산법능구유효지대병변엄중적CT도상진행복강분할，대조성화기관변형유량호적로봉성。
The organs with cancers in CT image deform seriously and the traditional segmentation methods cannot segment the entire chest correctly. This paper presents a 3D mesh based segmentation method. Firstly, an initial chest skeleton is found using ribs and back bones; secondly a sphere triangle mesh is constructed and the correspondence between the mesh and the chest skeleton is created; thirdly the mesh is deformed according to three objective functions; finally the mesh is refined by edge voxels near the mesh and the segmentation result is refined. Experimental results show that the proposed method can efficiently segment the seriously-deformed chest in CT image, which is robust against the noise and deformation of organs.