摘要
目的:提出一种基于Hessian矩阵及余弦定理的血管检测方法,改善肝门静脉血管检测效果。方法:充分考虑肝脏内肝门静脉CT图像的灰度及结构特点,首先利用肝脏图像的低灰度阈值对CT图像进行预处理,然后提出由余弦定理构造新的血管相似性函数,并计算多尺度的函数响应值,最后利用肝门静脉图像的低灰度阈值和连通域提取方法提取最大连通域。结果:该方法能有效地检测肝门静脉血管,检测结果与IRCADb数据集中专家的人工分割结果更接近。结论:基于Hessian矩阵及余弦定理的肝门静脉血管检测方法为后续的分割和三维重建奠定了良好的基础。
Objective To improve the detection of the portal vein on computed tomography （CT） images with a new vessel detection method based on Hessian matrix and the cosine law.Methods By fully considering the gray level and structural characteristics of the portal vein on CT images,we used a low gray threshold of the liver for preprocessing of the input CT images.We proposed a new vascular similarity function based on the cosine law to compute multi-scale function response,and utilized the low gray threshold of the portal vein image and the connected domain extraction method to extract the maximum connected domain.Results The proposed vessel detection method achieved the effective detection of the portal vein on CT images,and the detection results was closer to the manual results made by doctor in IRCADb database.Conclusion The vessel detection method based on Hessian matrix and cosine law can facilitate subsequent image segmentation and three-dimensional reconstruction.
作者
邓金城
刘常春
莫珍丽
陈子杰
吴江华
DENG Jincheng LIU Changchun MO Zhenli CHEN Zijie WU Jianghua(Shenzhen Yino Intelligence Techonology Co.,Ltd., Shenzhen 518057, China Shenzhen Mindray Bio-medical Electronics Co., Ltd., Shenzhen 518057, China)
出处
《中国医学物理学杂志》
CSCD
2017年第5期462-467,共6页
Chinese Journal of Medical Physics
基金
深圳市高技术产业化扶持计划(股权资助)(S2016I65100017)
关键词
HESSIAN矩阵
余弦定理
肝门静脉
CT图像
图像增强
血管检测
Hessian matrix
cosine law
hepatic portal vein
computed tomography image
image enhancement
vessel detection
