欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
肺癌是人类的一大杀手,为了提高其治愈率,人们越来越重视对肺癌的早期形态——肺结节的影像检测,但一直被较高的假阳性率所困扰。在高分辨率CT图像基础上,打破常规思维,从肺部血管三维重建入手,间接去掉血管组织对结节提取的干扰。首先利用数学形态学及凸包算法获得二维完整肺实质,再利用区域增长法提取肺部软组织,间接得到疑似结节图像,然后利用三维Hessian矩阵特征值的几何意义,构造三维血管结构的增强因子,得到完整的肺部血管图像,将其与疑似结节图像进行对比,重合区域即可除去,最大限度地剔除血管的干扰,最后再利用疑似区域的几何特征剔除残余的肺部杂质,最终获得较低的假阳性率,提取准确率较高。
英文摘要：
Lung nodules are one of the most common lesions,thus early detection and diagnosis of lung nodule is critical for the medical treatment of lung cancer.However,the false positive is mostly high with general methods.We proposed to begin with 3D reconstruction of vessels in lung to avoid the interference indirectly.First,we got the 2D complete lung regions.Second,the lung soft tissue was acquired by region growing method to get the possible nodules.Third,the geometrical meaning of 3D Hessian matrix′s eigenvalues was used to finish vessel 3D enhancement and reconstruction.Then the vessels image was compared with the possible nodule image to remove the overcast regions.Finally,we removed the small lung regions with the geometry features.The proposed method was approved to be effective in getting low false positive and high accuracy.
同期刊论文项目
同项目期刊论文
Robust propagation velocity estimation of gastric electrical activity by least mean p-norm blind cha
期刊信息
位置：