%0 Journal Article
%T 基于胸部CT影像的肺血管树分割关键技术研究<br>Research on Key Technologies of Pulmonary Vascular Trees Segmentation Based on Thoracic CT Images
%A 杨志永
%A 肖洪旭
%A 李雨泽
%A 姜海松
%A 姜杉
%J 天津大学学报（自然科学与工程技术版）
%D 2018
%R 10.11784/tdxbz201706043
%X 肺支气管的排除和血管组织的精确探测, 是影响肺血管树分割精度的重要因素．经形态学处理后的CT影像可提高对器官信息的探测能力, 因此提出形态学辅助的区域生长方法用于支气管分割, 并引入泄漏判断条件抑制分割泄漏．针对血管组织的提取, 提出多阈值分割方法, 通过引入多尺度滤波器获取不同尺寸半径血管的最大响应尺度信息, 计算血管组织相应的分割阈值, 实现分割阈值的动态匹配．实验结果表明：应用于10套CT影像, 血管组织分割准确率为97.062% , 血管分支抽取率为93.95% , 肺血管树分割精度得到较大提高．<br>The elimination of lung bronchus and the precise detection of vascular tissues are the major factors affecting the segmentation accuracy of pulmonary vascular trees. The morphological disposal with CT image could improve the ability of organ detection，thus a morphology-assisted region growing method was proposed to segment bronchus. Besides，the leak judgment function was also introduced to avoid the leakage phenomenon. In order to extract vascular tissues，the multi-threshold segmentation method was proposed，which is based on the multi-scale enhancement filter. By acquiring the max response information of vascular tissues with different radius，the corresponding segmentation threshold of vascular tissues was calculated and the dynamic matching between vascular tissues and segmentation threshold was achieved. Being applied to 10 sets of CT images，the proposed algorithm exhibited promising results. The segmentation accuracy rate of vascular tissues and the extraction rate of vascular branches were 97.062% and 93.95% ，respectively，considerably improving the segmentation accuracy of pulmonary vascular trees
%K 肺血管树
%K 胸部CT影像
%K 区域生长
%K 多阈值分割&lt
%K br&gt
%K pulmonary vascular trees
%K thoracic CT images
%K region growing
%K multi-threshold segmentation
%U http://journals.tju.edu.cn/zrb/oa/darticle.aspx?type=view&id=201802009