摘要利用计算机进行医疗辅助诊断近年来逐渐成为一个热门的应用。本文以腺体癌细胞的研究为背景，进行辅助诊断系统的研究。腺体癌细胞的研究利用组织形态学，用4C规则把腺体细胞分成4类。4C规则表示测量细胞的4个参数，这4个参数一定概率上反应了细胞的癌情况。通过该4个参数的测量，本文使用BP神经网络对其进行分类。作为比较，还使用了D评分诊断方法与两种方法的综合，比较它们之间的效果。63966

本文共分为五部分，第一章对研究背景及研究内容作了简要的介绍；第二章介绍了数字图像处理和神经网络的相关理论；第三章分析了系统的需求；第四章着重介绍了系统的实现，包括参数的测量过程和用样本集训练BP神经网络，验证效果；最后一章本文作了总结和对系统的未来作了展望。

关键词 医疗辅助诊断 BP神经网络 4C规则 D评分

毕业论文 外 文 摘 要

Title  Design and Realization of Aided Diagnosis Medical Image Processing System

Abstract Computer aided diagnosis in recent year has become a popular application. With the research of gland cancer as its background, this paper studied the aided diagnosis system. Through the histomorphometry study, the gland cells are classified into four types by 4C-roles. 4C-roles indicates the four parameter of the measured gland cells, which reflect the cell situations in certain probability. By measuring these four parameters, this paper uses the BP neural network to classify them. For comparison, we also used the D-score diagnostic method and integrated two method to compare the results between them.

This paper falls info five chapters. In the first chapter, the research background and content is simply discussed. In the second chapter, the theory of digital processing and neural network is introduced. In the third chapter, analysis the system requirements. In the four chapters, system realization, including the process of measuring parameter and the training BP neural network with sample set, is introduced in detail. In the last chapter, it summarizes the paper and forecasted the future development of this system.

Keywords  computer-aided medical diagnosis , BP neural network , 4C-roles, D-score

目  录

1 绪论 1

1.1 研究背景 1

1.2 研究目的 2

1.3 本文研究的主要内容、方法和思路 2

1.3.1 研究内容 2

1.3.2 研究方法及思路 3

2 数字图像处理及神经网络 5

2.1 数字图像处理 5

2.1.1 数字图像处理 5

2.1.2 数字图像处理方法 6

2.1.3 Image Pro Plus图像处理 7

2.2 神经网络 7

2.2.1 神经网络 7

2.2.2 神经网络相关模型 9

3辅助诊断过程分析 13

3.1诊断概述 13

3.1.1 诊断内容 13

3.1.2 诊断目的 14

3.2诊断过程概述 14

3.2.1 诊断前的准备 14

3.2.1 诊断结果处理 15

4 系统实现 16

4.1 医学图像参数的获取