基于局部Chan-Vese模型的超声颈动脉图像水平集分割.docVIP

  基于局部 Chan-Vese 模型的超声颈动脉图 像水平集分割# 曾雅洁1，黄金河1，杨鑫2，丁明跃1** 5 10 15 20 25 30 35 40 （1. 华中科技大学生命科学与技术学院，生物医学工程系，图像信息处理与智能控制教育 部重点实验室，武汉 430074； 2. 华中科技大学图像识别与人工智能研究所，多谱信息处理技术国防科技重点实验室， 武汉 430074） 摘要：心血管疾病是世界上三大致死疾病之一。动脉粥样硬化会引发多种心脑血管疾病，动 脉粥样硬化斑块是其最主要的病理形态学改变。本文对超声主颈动脉（CCA，Common Carotid Artery）横向图像中血管的内外膜进行分割，结果可用于对斑块大小、厚度和形状的定性估 计及定量测量。本文选用局部 C-V(LCV，Local Chan-Vese)模型分割外膜；C-V 模型分割内 膜；引入内外膜距离限制项限制内膜的演化，并用稀疏场方法（SFM，Sparse Field Method） 提高水平集算法的效率；通过全正交法（FOM，Full-Orthogonal Method）、射线法、相似系 数分析法等对结果进行分析比较。实验结果表明，LCV 模型可有效地分割颈动脉血管外膜， C-V 模型可以有效地分割血管内膜，改进方法提升了程序运行速度，提高了内膜的分割精度。 关键词：生物医学工程；图像分割；颈动脉超声图像；水平集；局部 C-V 模型；稀疏场算 法 中图分类号：TP391.4 Ultrasound carotid image segmentation using level set method based on local Chan-Vese model ZENG Yajie1, HUANG Jinhe1, YANG Xin2, DING Mingyue1 (1. Image Processing and Intelligence Control Key Laboratory of Education Ministry of China,School of Life Science and Technology,Huazhong University of Science and Technology, WuHan 430074; 2. State Key Laboratory for Multi-spectral Information Processing Technologies, Institute for Pattern Recognition and Artificial Intelligence(IPRAI),Huazhong University of Science and Technology, WuHan 430074) Abstract: Cardiovascular disease is one of the three major killer diseases in the world. Atherosclerosis will lead to a variety of cardiovascular and cerebrovascular diseases. Atherosclerotic plaques is the major pathological changes of atherosclerotic.The segmentation of intima and adventitia of Common Carotid Artery(CCA) is the main work, and the result can be used for qualitative estimates and measurements of plaque size, thickness and shape. The paper chose Local C-V model to segment the adventitia and C-V model to segment the intima.Distance Limitations Item was proposed to limit the evolution of the intima,and Sparse Field Method(SFM) was used to improve the efficiency of the level set method.The result was analyzed and compared by full-orthogonal method (FOM), ray method and Dice index.The results