基于改进C-V模型的血管内膜超声图像分割  

Segmentation of Vascular Intima Ultrasound Images Based on an Improved C-V Model

作　　者：胡玉洁[1] 蒋玲[2] 宋海波[3] 刘奇[1] 

机构地区：[1]四川大学电气信息学院,成都610065 [2]成都大学实验技术中心,成都610106 [3]四川大学华西医院超声科,成都610041

出　　处：《生物医学工程学杂志》2010年第5期1115-1119,共5页Journal of Biomedical Engineering

基　　金：国家自然科学基金生命科学部资助项目(30700781)

摘　　要：血管一定程度上能反映心血管的病变。要快速准确地对血管内膜超声图像进行分割,在经典C-V模型基础上,引入惩罚项,避免水平集函数演化过程中的重初始化,并使曲线演化受控于区域外能项,加快分割速度。计算边界点位移变化,讨论其代表的生理病理意义。改进算法能较高效准确地提取血管内膜超声图像目标轮廓。分析边界点的相对位移关系曲线,不仅能反映边界的提取效果,还可一定程度上反映血管或心律的异常。In view of the fact that blood vessels can reflect, to a certain degree,the pathologic changes of cardiovascular diseases,research workers count it necessary to take quick and accurate segmentation of neointimal ultrasound images.The segmentation is faster owing to the reconciled signed distance function added into the classical C-V mode,thus avoiding the step of re-initialization.Moreover,the curve evolution is controlled by external energy item.The displacement of boundary points is calculated.The result shows that the improved model can effectively and accurately extract the contour of neointimal ultrasound images.The curve can not only assess the extracted boundary but also reflect the abnormality of vascular or cardiac rhythm.

关 键 词：血管内膜超声图像 活动轮廓 C-V模型 重初始化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]