计算机图像处理技术辅助精子运动能力分析研究

【作者】 朱安定；

【导师】 夏顺仁；

【作者基本信息】 浙江大学 ， 生物医学工程， 2002， 硕士

【摘要】 计算机辅助精子运动能力自动分析技术是计算机图像处理技术在医学辅助诊断上的典型应用。传统的人工肉眼观察计数法费时费力，而且精度难以保障。随着微型计算机技术的飞速发展，使基于微型计算机的低配置辅助诊断系统成为可能。利用计算机图像分析技术不仅分析速度快，而且计算精度高、重复性好。所以计算机辅助分析已经成为现代生殖实验室中对男性不孕症诊断和人工授精样本优选的先进手段和追求目标。为此，我们开展了计算机图像处理技术辅助精子运动能力分析研究，取得了满意的结果。 本论文在给出系统实现的系统结构基础之上，详细阐述了系统的设计原理，提出了我们自己的帧内分割和帧间跟踪的精子动态图像分析的思想，给出了所有量化参数计算的全部算法流程，同时对分析精度的影响因素、分析质量的评估方法、采样频率对参数计算的影响等方面进行了深入的讨论。 针对精子显微图像的特点，论文分析比较了四种现存的图像分割算法的特点，进而判定：基于灰度直方图的最大类间方差阈值选取法是最适合不同背景光强和样本密度的分割方法。 论文还分析了不同的多目标跟踪算法，给出了在特定采样频率条件下能够跟踪的目标速度和样本密度之间应该满足的条件公式，提出了适于不同条件的多因素综合的邻域匹配跟踪算法。 24例临床样本比对研究结果表明，我们研究的算法能够满足临床的应用需求，有着广泛的推广价值和应用前景。更多还原

【Abstract】 Computer assisted sperm motion analysis system (CASMA) is a typical application of computer image processing technology in computer assisted diagnosis (CAD). The classical manual method cannot reach an accurate result efficiently. The booming of microcomputer technology makes low-cost configurable CASMA possible. Using the computer technology, CASMA system can reach not only a high precision but also a sound reproducibiliry. Nowadays the CASMA system has become an advanced tool for male sterility in modern andrology and reproductive laboratory.At first the system structures and the analysis theories are described in details. Then the new idea of using two steps, that is to say, inner-frame image segmentation and inter-frame multitarget tracking to analysis the sequence frames is presented. And all the algorithms of parameters extracting are presented too. Otherwise, the factors, which are affecting system precision, the evolution methods to analysis quality and several different aspects of system are discussed deeply.In this paper, according to the spermatozoa microscopic image characteristics four conventional image segmentation algorithms are compared. And at last the maximum between-class variance method is determined to be the best segmentation algorithm to the special images.In this paper, several different multitarget tracking algorithms are also compared, and the relationship between the maximum velocity and the concentration under the special image capture frequency is brought out. A novel tracking strategy of referring to the velocity and direction of the sperm head based on the neighbor matching technology is introduced in our algorithm.The comparison results of 24 specimens using our algorithms and an American product HTM-IVOS system shows that our algorithms are valuable for the clinical analysis and have a bright application prospect.更多还原