Demons算法胸片图像自动配准算法的初步研究
本文针对胸片图像配准，首先进行了图像去噪、增强、以及分割的图像预处理，然后尝试使用了光流法和传统的Demons方法，结果表明并不能获得令人满意的配准结果。在此基础上，本文提出了一种改进的Demons算法，通过在传统的Demons方法中引入来自目标图像的驱动内力进行改进。通过采用多种评价方法进行定性与定量的分析，结果表明改进的Demons方法优于光流法和传统的Demons算法，对胸片图像的配准准确有效。
关键词 胸片 图像配准 Demons算法
毕业论文设计说明书外文摘要
Title Preliminary Study on Automated Registration Algorithm for Chest Radiographs
Abstract
Chest radiographs have been widely used for detection and diagnosis of thoracic disease. However, observing the difference between chest radiographs is rather difficult due to patients’ respiratory motion and various photographic conditions. Image registration aims to establish the correspondence between images. Therefore, it is of great significance to develop an automated and accurate image registration algorithm to enhance the change between chest radiographs.
In this study, image preprocessing is initially carried out for noise reduction, image enhancement and segmentation. Following that, Optical flow and conventional Demons methods are investigated and applied to align chest radiographs. The preliminary results show that neither one leads to the reasonable results. According to the characteristics of chest radiographs, We propose an improved Demons algorithm by modifying the conventional Demons registration metric with a driving force from the target image. By qualitatively and quantitatively validating the registration using various methods, results show that the improved Demons method obtains effect and efficient registration of chest radiographs, which is superior to the optical flow and conventional Demons methods.
Keywords Chest radiograph Image registration Demons algorithm
目次
1 绪论 1
1.1 研究背景及意义 1
1.2 国内外研究及现状 3
1.3 本文研究内容与结构安排 5
2 数据来源与图像预处理 6
2.1 本文数据来源与实验环境 6
2.2图像预处理 6
2.3 本章小结 14
3 图像配准流程、原理及研究 15
3.1光流法 15
3.2传统的Demons方法 17
3.3改进的Demons方法 19
3.4 本章小结 21
4 评价方法与结果分析 22
4.1 配准效率分析 22
4.2 五分类法分析 22
4.3 直方图宽度法分析 24
4.4 变形场分析 26
4.5 本章小结 27
5 总结与展望 28
5.1 总结 28
5.2 展望 28
致 谢 29
参考文献30
1 绪论
1.1 研究背景及意义
近年来，医学影像的发展为临床诊断和治疗提供了有效的辅助手段，特别是随着各种新的医学成像技术的发展和在临床中的广泛应用，医学影像不只应用于通常的疾病诊断之中，更是遍布整个临床活动的周期，医生需要多次结合医学图像，完成不同阶段的治疗。胸透成像是医学影像中最简单、最广泛使用的一种X射线投影技术。