癌细胞DNA含量测量中的图像处理研究

3.0 侯斌 2024-11-19 4 4 2.61MB 66 页 15积分

侵权投诉

摘要

近些年来，癌症患者的数量呈上升趋势,使得癌症及肿瘤科学的研究异常紧迫。

肿瘤细胞 DNA 含量的测定和分析对恶性肿瘤的早期病理诊断、恶性程度判定、疗

效估价和预测预后具有重要价值。传统的方法是采用流式细胞仪(Flow Cytometer)，

虽然它有精确度高、速度快、检测细胞数量多和能进行细胞周期分析等优点，但

是细胞的形态特征在处理过程中会被丢失，且设备复杂，仪器价格昂贵，难以广

泛应用。而如果应用图像光度术（Image Cytometer）则不仅能对组织细胞切片上

极少量的细胞核作 DNA 含量测量和倍体分析，同时可以测量对诊断和预后判断都

极具价值的细胞核的某些形态参数。

应用图像光密度术检测病理切片时，病理切片细胞图像中的噪声、不均匀背景

和粘连细胞，将严重影响 DNA 含量测定和分析的准确性。因此，图像处理技术在

癌细胞 DNA 含量测量中有着十分重要的意义。

本文主要应用 ICM 方式来测定和分析肿瘤细胞的 DNA 含量，研究了如何从图

像处理的方面，如图像的去噪、分割、粘连细胞的分离、细胞的特征提取及 DNA

含量的计算等方面进行改进，提高肿瘤细胞 DNA 含量测定的正确性与效率。

在图像预处理中，本文首先介绍了图像二值化、图像去噪与图像增强的原理，

并提出了对于背景不均匀细胞图像的二值化方法。接着对比了三种自适应阈值分

割算法的分割效果及耗用时间，包括最佳阈值法、最大类间方差法和最佳熵阈值

法，选择出最适合本文癌细胞 DNA 染色图像的分割方法。最后，对已失去形态特

征意义的边缘细胞，应用了区域生长的方法进行边缘像素擦除。

在粘连细胞的分离中，本文首先简单描述了腐蚀膨胀算法和测地重建算法的

缺点。接着，详细阐述了分水岭算法的三种实现过程，指出对于形态异型性较大

的细胞显微图像中等值淹没更为合理，并在现有算法的基础上，提出了新的基于

地形图的等值跟踪算法，也就粘连分离效果及耗用时间对三种方法进行了性能比

较；同时，还详细描述了陆宗骐教授的链码差法的实现过程，提出了有效的凹点

配对原则――“相亲原则”，就性能与前三种分水岭实现方法进行了比较。最后简

单带过三种其它算法：矢量夹角法、估算圆心法和多边形近似法。

在细胞的特征提取中，本文首先从链码差法的粘连分离结果图像中，介绍了

单细胞的周长、圆心、面积和圆度的计算方法。之后，估算了现有图像中细胞的

DNA 含量。

关键词：DNA 含量粘连分离分水岭链码跟踪图像分割

III

ABSTRACT

Recently, the upward trend of the cancer patient number makes it necessary to

study on cancer and knub. Identification and analysis of DNA content in tumor cells

have great value on early pathological diagnose 、malignant degree estimation 、

treatment evaluation and prediction of malignant tumor. The traditional method is

using Flow Cytometer, which is precise, fast, can work on large scale and can analysis

cell cycle, but the morphological characters would be lost in the procedure, and the

complex and expensive equpiments also made this method hard to widely apply.

While Image Cytometer(ICM) can measure the DNA content of cell nucleus, analysis

the chromosome number based on farthing cells in tissue, further more it can measure

some morphological characters which are of great importance for diagnosing and of

prediction of cancer.

When using ICM to measure the pathology slice cell, the noise, non-uniform

background and adhesion cells on the image would seriously infect the correctness of

measurement and therefore the analysis of DNA content. Consequently, image process

technique has key important meaning on measurement of DNA content on cancer

cells.

This paper mainly uses ICM method to measure and analysis the DNA content of

cancer cells, studies the method to improve the correctness and efficiency through

image process, including denoise, partition, separation of adhesion cells, character

distill and DNA content calculation.

During the preprocess step, this paper firstly introduces the principle of image

thresholding, image denoising, and image enhancement. Then it raises a threshold

method to cell image with non-uniform background. Furthermore it compares the

threshold effect and time consuming of three adaptive threshold algorithms including

Best Threshold Method, Otsu Method and KSW Entropic Method, and chooses the

most appropriate threshold method to the cancer cell DNA chromatic image in this

paper. Finally, the edge cells that have lost the morphological significant is removed

by region growing method.

In the adhesion cells separation step, this paper simply describes the weakness of

Erosion Dilation Algorithm and Distance Image Reconstruction Method in the

beginning. Then it elaborates three achievements of Watershed Method, and indicates

that for cell micro image with large morphological difference, it's more reasonable to

Equivalent submerge. Based on the existing algorithms, it raises a new Trace and

Mark Algorithm Based on Distance Transformation and compares the adhesion cell

separation efficiency and time consuming of three achievements. At the same time, it

describes the achievement on chain code difference by Professor Lu, Zongqi in detail,

which proposes the effective concave pair principle----Dating Principle, and also

compares its efficiency with other three achievements. Finally it briefly tells the other

three algorithms (Vector Angle Method, Circle Center Estimation Method, and

Optimal Polygon Approximation Method)

During the cell character distill step, this paper starts with adhesion cell

separation result image by Chain Code Difference Method, followed by introduction

the calculation method of perimeter, center, area, and roundness. Finally this paper

estimates the DNA content of the existing image cell.

Key word: DNA content; adhesion separation; watershed; code

tracing; image segmentation

摘要

ABSTRACT

第一章绪论 ..............................................................................................................1

§1.1 课题背景 ....................................................................................................... 1

§1.2 DNA 含量测定的发展现状 ........................................................................... 1

§1.2.1 流式细胞术 ....................................................................................... 1

§1.2.2 图像细胞光度术 ............................................................................... 2

§1.3 课题选题意义和依据 ................................................................................... 3

§1.4 论文主要研究内容及结构 .......................................................................... 3

第二章 ICM 定量测定技术的基本原理 .................................................................. 4

§2.1 细胞的分裂周期 ........................................................................................... 4

§2.2 肿瘤 ............................................................................................................... 5

§2.3 DNA 含量测定原理 ....................................................................................... 5

§2.3 本章小结 ....................................................................................................... 7

第三章图像的预处理 ..............................................................................................8

§3.1 图像的获取 ................................................................................................... 8

§3.1.1 标本处理 ............................................................................................ 8

§3.1.2 显微图像采集系统 ............................................................................ 8

§3.2 灰度直方图 .................................................................................................. 9

§3.3 自适应阈值分割 ........................................................................................ 10

§3.3.1 最佳阈值法 ...................................................................................... 11

§3.3.2 最大类间差法 .................................................................................. 12

§3.3.3 最佳熵阈值法(KSW 熵法) ............................................................... 13

§3.3.4 三种阈值分割方法比较 ................................................................. 14

§3.4 图像去噪与增强 ........................................................................................ 17

§3.4.1 图像的平滑 ...................................................................................... 17

§3.4.2 图像的锐化 ...................................................................................... 18

§3.5 擦除边缘细胞 ............................................................................................ 21

§3.5.1 区域生长 ......................................................................................... 21

§3.6 本章小结 ..................................................................................................... 22

第四章细胞图像的粘连分离 ................................................................................24

§4.1 腐蚀膨胀法 ................................................................................................ 24

§4.1.1 算法原理 ......................................................................................... 24

§4.1.2 算法优缺点 ..................................................................................... 25

§4.2 测地重建法 ................................................................................................ 26

§4.2.1 算法的原理 ..................................................................................... 26

§4.2.2 算法的优缺点 ................................................................................. 26

§4.3 分水岭法 ..................................................................................................... 27

§4.3.1 算法的原理 ...................................................................................... 27

§4.3.2 算法实现方法一 ............................................................................. 27

§4.3.3 算法实现方法二 ............................................................................. 31

§4.3.4 算法实现方法三 ............................................................................. 33

§4.3.5 三种实现方法比较 .......................................................................... 35

§4.4 链码差法 ..................................................................................................... 38

§4.4.1 算法的原理 ..................................................................................... 38

§4.4.2 算法的优缺点 ................................................................................. 44

§4.5 其它算法 .................................................................................................... 46

§4.5.1 矢量夹角法 ..................................................................................... 46

§4.5.2 估算圆心法 ..................................................................................... 47

§4.5.3 多边形近似法 ................................................................................. 47

§4.6 本章小结 ..................................................................................................... 48

第五章癌细胞的特征提取 ....................................................................................50

§5.1 形态特征提取 ............................................................................................ 50

§5.1.1 周长 ................................................................................................. 51

§5.1.2 圆心 ................................................................................................. 51

§5.1.3 面积 ................................................................................................. 51

§5.1.4 圆度 ................................................................................................. 53

§5.2 DNA 含量统计分析 ..................................................................................... 53

§5.3 本章小结 .................................................................................................... 55

第六章总结与展望 ................................................................................................56

总结 ..................................................................................................................... 56

展望 ..................................................................................................................... 57

参考文献 ..................................................................................................................... 58

在读期间公开发表的论文和承担科研项目及取得成果 ......................................... 61

致谢 ......................................................................................................................... 62

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

15 积分 4人已下载

立即下载

摘要：

I摘要近些年来，癌症患者的数量呈上升趋势,使得癌症及肿瘤科学的研究异常紧迫。肿瘤细胞DNA含量的测定和分析对恶性肿瘤的早期病理诊断、恶性程度判定、疗效估价和预测预后具有重要价值。传统的方法是采用流式细胞仪(FlowCytometer)，虽然它有精确度高、速度快、检测细胞数量多和能进行细胞周期分析等优点，但是细胞的形态特征在处理过程中会被丢失，且设备复杂，仪器价格昂贵，难以广泛应用。而如果应用图像光度术（ImageCytometer）则不仅能对组织细胞切片上极少量的细胞核作DNA含量测量和倍体分析，同时可以测量对诊断和预后判断都极具价值的细胞核的某些形态参数。应用图像光密度术检测病理切片时，病理...

展开>> 收起<<

癌细胞DNA含量测量中的图像处理研究.pdf

共66页,预览7页

还剩页未读，继续阅读

癌细胞DNA含量测量中的图像处理研究

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

推荐作者

热门标签

举报选择: