Snake 法和GVF snake 法进行颈动脉超声图像分割的比较
金娇英1，杨鑫2，邱武1，丁明跃1
基金项目：高等学校博士学科点专项科研基金（20070487056），中国博士后科研基金（20100480906）
作者简介：金娇英（1989-），女，本科，医学图像处理
通信联系人：丁明跃（1961-），男，教授，医学影像，三维医学超声成像，图像导引手术与治疗，电阻抗成像技术研究，计算机视觉，三维重建，航迹规划，图像天候模型研究.
（1. 华中科技大学生命科学与技术学院，武汉 430074；2. 华中科技大学模式识别与人工智能研究所，武汉 430074）
摘要：目的：针对颈动脉超声图像，分析比较运用经典Snake 和GVF snake 确定颈动脉的内外轮廓边界，探讨适合颈动脉超声图像分割的方法。方法：将经典Snake 和GVF snake 作为分割工具，对38 组颈动脉超声图像进行内外轮廓分割，实验过程中，根据分割的结果，不断地对两个模型的参数进行调整，使得两种分割方法的效果分别达到最优，最后比较两种分割方法的分割结果，选择收敛速度快、分割效果准确的方法。结果：当初始化轮廓非常贴近目标轮廓时，经典Snake 和GVF snake 均能获得较好的分割结果，但GVF snake 的收敛速度快于经典Snake；当初始化轮廓距离目标轮廓较远时，经典Snake 分割失败，而GVF snake仍可以得到准确的收敛结果。结论：对于简单的图像，经典Snake 和GVF snake 均能得到较好的分割结果；对于较复杂及边界凹陷的图像，GVF snake 的分割效果明显优于经典Snake。
关键词：图像分割； Snake； GVF snake； 颈动脉超声图
中图分类号：TP391.41
The comparison between the Snake and the GVF Snake in carotid ultrasound image segmentation
Jin Jiaoying1, Yang Xin2, Qiu Wu1, Ding mingyue1
(1. School of Life Science and Technology,Huazhong University of Science and Technology,Wuhan 430074;2. Institute of Pattern Recognition and Artificial Intelligence ,Huazhong University of Science and Technology, Wuhan 430074)

 Abstract: Objective Traditional snake and GVF snake are used to determine the internal and external contours of carotid ultrasound images. Compared the results of these two methods, an effective method is set up for the carotid ultrasound images segmentation. Methods Traditional snake and GVF snake were used to segment 38 groups of carotid ultrasound images. The parameters of these two models were adjusted to get optimal segmentation results. The segmentation results were compared. Results Both the traditional snake and the GVF snake resulted in satisfactory results when the initial contours
were close to the contours of the objects, but the convergence rate of the GVF snake was faster than that of the traditional snake. The GVF snake also brought satisfactory results when the initial contours were far from the contours of the objects while the traditional snake failed. Conclusion Both the traditional snake and the GVF snake can result in satisfactory results in simple images segmentation. The GVF snake does better than the traditional snake in the segmentation of complex images with concaves.
Key words: image segmentation; snake; GVF snake; carotid ultrasound images